Internet of vehicles service quality warning method and device, storage medium and product
By using a hierarchical detection model to perform hierarchical detection and anomaly alarms on vehicle network data, the problems of missed fault detection and poor data coupling in existing technologies are solved, and higher detection accuracy and operation and maintenance efficiency are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA MOBILE GRP BEIJING
- Filing Date
- 2025-12-11
- Publication Date
- 2026-04-24
AI Technical Summary
Existing methods for identifying and alarming the quality of connected vehicle services suffer from problems such as missed fault detection due to single-index data detection and poor coupling between various service data, resulting in low detection accuracy and difficulty in effectively carrying out network operation and maintenance.
A hierarchical detection model is adopted to perform hierarchical detection on vehicle network data. By preprocessing historical network data and generating datasets with multiple aggregation levels, different aggregation levels correspond to business datasets with different spatiotemporal dimensions. Anomaly detection is performed using the hierarchical detection model, and alarms are suppressed and integrated according to the priority of aggregation level from large to small.
It reduces missed and incorrect fault diagnosis, improves detection accuracy, and enhances the fault resolution efficiency of the operations and maintenance team, enabling effective network operations and maintenance work.
Smart Images
Figure CN121924508A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of Internet of Things (IoT) technology, and in particular to a method, device, storage medium, and product for alarming the quality of vehicle-to-everything (V2X) services. Background Technology
[0002] In the 5G private network vehicle-to-everything (V2X) industry, network quality is closely related to V2X applications (such as remote vehicle control, in-vehicle entertainment, in-vehicle navigation, voice and SMS communication, etc.). Network anomalies often cause vehicle malfunctions or even prevent vehicles from starting. To ensure normal vehicle operation, operators typically organize relevant maintenance teams to maintain and support the network.
[0003] However, existing methods for identifying and alarming the quality of connected vehicle services suffer from problems such as missed fault detection due to single-index data detection and poor coupling between various service data, resulting in low detection accuracy. This makes it difficult for the operation and maintenance team to effectively carry out network operation and maintenance work. Summary of the Invention
[0004] To address the aforementioned issues, this disclosure provides a method, device, storage medium, and product for issuing alarms for vehicle-to-everything (V2X) service quality.
[0005] According to a first aspect of the present disclosure, a method for issuing vehicle-to-everything (V2X) service quality alarms is provided, the method comprising: Historical network data of the Internet of Vehicles within a preset time range is preprocessed and aggregated according to multiple aggregation levels required by the business to obtain the datasets corresponding to each aggregation level. Different aggregation levels correspond to business datasets with different spatiotemporal dimensions. The datasets at the multiple aggregation levels are input into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset. Different aggregation levels correspond to different levels of detection sub-models in the hierarchical detection model. Based on the anomaly detection results of the datasets at each aggregation level, alarms are suppressed and integrated according to the priority of aggregation level from large to small, and alarm information is output.
[0006] In one implementation, the step of preprocessing historical network data of the vehicle network within a preset time range and aggregating it according to multiple aggregation levels required by the business to obtain datasets corresponding to each aggregation level includes: The historical network data is processed for missing and outlier values. And aggregate according to multiple aggregation levels required by the business to obtain candidate aggregate datasets; Candidate aggregated datasets with a sample size below a set threshold are filtered out to obtain the datasets corresponding to each aggregation level; The multiple aggregation levels include a first-level aggregation, a second-level aggregation, a third-level aggregation, and a fourth-level aggregation. The first-level aggregation is used to aggregate spatiotemporal service datasets at the vehicle level; the second-level aggregation is used to aggregate spatiotemporal service datasets at the data network name (DNN) level; the third-level aggregation is used to aggregate spatiotemporal service datasets at the cell level; and the fourth-level aggregation is used to aggregate spatiotemporal service datasets at the district / county level. The vehicle level is lower than the DNN level, the DNN level is lower than the cell level, and the cell level is lower than the district / county level.
[0007] In one implementation, the hierarchical detection model includes a first-level anomaly detection sub-model corresponding to a first-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset includes: Using the first-level anomaly detection sub-model, the baseline mean and first covariance matrix of each indicator in the vehicle-level spatiotemporal business dataset are calculated using the minimum covariance determinant algorithm. Based on the baseline mean and the first covariance matrix, calculate the first Mahalanobis distance between each vehicle's index vector and the corresponding baseline mean; If the first Mahalanobis distance is greater than a first preset threshold, the vehicle is determined to be abnormal. If the vehicle is abnormal and there are no higher-level abnormalities, output the first list of abnormal indicators.
[0008] In one implementation, the hierarchical detection model includes a second-level anomaly detection sub-model corresponding to the second-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset includes: The second-level anomaly detection sub-model calculates the baseline values of each indicator and the second covariance matrix of the spatiotemporal business dataset of the specified region within the same time period in the past specific time window using the weighted moving average; Based on the baseline values of each indicator and the second covariance matrix, the deviation between each indicator vector of the spatiotemporal business dataset at the DNN level and the corresponding baseline value is calculated for the current time period, and if the deviation is greater than the second preset threshold, the specified DNN is determined to be abnormal in the current time period. If the specified DNN is anomaly, and there are no higher-level anomalies, output a second list of anomaly indicators; or, Calculate the second Mahalanobis distance between the specified DNN and other DNNs within the specified region in the multi-index space of the spatiotemporal business dataset at the DNN level, and perform a horizontal comparison between DNNs by comparing the second Mahalanobis distance with a third preset threshold determined based on the chi-square distribution, in order to determine whether there are any spatially isolated DNN anomalies. If the second Mahalanobis distance is greater than the third preset threshold, the specified DNN is determined to be abnormal; If the specified DNN is abnormal and there are no higher-level abnormalities, output a third list of abnormality indicators.
[0009] In one implementation, the hierarchical detection model includes a third-level anomaly detection sub-model corresponding to the third-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset includes: Using the third-level anomaly detection sub-model, based on historical network indicator data, the current theoretical values of each indicator in the spatiotemporal service data of the specified cell are predicted using an exponential smoothing model, and the prediction deviation is calculated. The prediction deviation of at least one target indicator is continuously monitored using a cumulative sum algorithm. When the cumulative deviation exceeds a fourth preset threshold, the designated cell is determined to be a persistent anomaly in the time series. If the specified cell is abnormal, and there are no higher-level abnormalities, output a fourth list of abnormal indicators; or, Calculate the differences of specified indicators between the specified cell and multiple physically neighboring cells, and use a t-test to determine whether the differences of the specified indicators are significant, so as to determine whether the specified cell is a spatially isolated anomaly. If the specified cell is abnormal and there are no higher-level abnormalities, output the fifth list of abnormal indicators.
[0010] In one implementation, the hierarchical detection model includes a fourth-level anomaly detection sub-model corresponding to the fourth-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset includes: Using the fourth-level anomaly detection sub-model, the proportion of abnormal vehicles in each district and county area of the district and county-level spatiotemporal business dataset is statistically analyzed within the sliding time window. Calculate the dynamic threshold of the proportion of abnormal vehicles within the sliding time window, the dynamic threshold being determined based on the mean and standard deviation of the historical proportion of abnormal vehicles within the time window; If the proportion of abnormal vehicles in the current district or county exceeds the dynamic threshold, it is determined that a large-scale anomaly has occurred in the current district or county, and a sixth list of abnormal indicators is output.
[0011] In one implementation, the anomaly detection results based on the datasets at each aggregation level are used to suppress and integrate alarms according to the priority from largest to smallest aggregation level, and alarm information is output, including: Following the principle that high-aggregation-level anomaly alarms have higher priority than low-aggregation-level anomaly alarms, high-aggregation-level anomaly alarms are output first, the triggering of low-aggregation-level anomaly alarms is suppressed, and the standardized alarm information corresponding to the high-aggregation-level anomaly alarms is output.
[0012] In one embodiment, the method further includes: Collect fault events confirmed by operation and maintenance personnel, and based on the fault events, optimize the algorithm parameters involved in at least one anomaly detection sub-model in the hierarchical detection model.
[0013] According to a second aspect of the present disclosure, a vehicle-to-everything (V2X) service quality alarm device is provided, the device comprising: The aggregation module is used to preprocess historical network data of the Internet of Vehicles within a preset time range, and aggregate the data according to multiple aggregation levels required by the business to obtain the datasets corresponding to each aggregation level. Different aggregation levels correspond to business datasets with different spatiotemporal dimensions. The detection module is used to input the datasets of the multiple aggregation levels into the hierarchical detection model for anomaly detection, so as to obtain the anomaly detection results of each aggregation level dataset. Different aggregation levels correspond to different levels of detection sub-models in the hierarchical detection model. The output module is used to suppress and integrate alarms based on the anomaly detection results of the datasets at each aggregation level, according to the priority of the aggregation level from large to small, and output alarm information.
[0014] According to a third aspect of the present disclosure, an electronic device is provided, comprising: A memory on which computer programs are stored; A processor for executing the computer program in the memory to implement the steps of the method of any one of the first aspects.
[0015] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the method described in any of the first aspects.
[0016] According to a fifth aspect of the present disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method described in any one of the first aspects.
[0017] The embodiments disclosed herein can achieve the following beneficial effects: By adopting a layered detection method to perform layered detection on vehicle network data, the embodiments disclosed herein can reduce the failure to detect faults and the failure to detect faults, improve the detection accuracy, thereby enabling the operation and maintenance team to effectively carry out network operation and maintenance work and improve the efficiency of fault resolution.
[0018] Other features and advantages of this disclosure will be described in detail in the following detailed description section. Attached Figure Description
[0019] The accompanying drawings, which are included to provide a further understanding of this disclosure and form part of this disclosure, illustrate exemplary embodiments of the present disclosure and are used to explain the disclosure, but do not constitute an undue limitation of the disclosure. In the drawings: Figure 1 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0020] Figure 2 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0021] Figure 3 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0022] Figure 4 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0023] Figure 5 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0024] Figure 6 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0025] Figure 7 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0026] Figure 8 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0027] Figure 9 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0028] Figure 10 This is a flowchart of a method for issuing alarms for the quality of connected vehicle services according to an embodiment of this disclosure.
[0029] Figure 11This is a block diagram of a vehicle network service quality alarm device provided in one embodiment of the present disclosure.
[0030] Figure 12 This is a block diagram of an electronic device provided in one embodiment of the present disclosure. Detailed Implementation
[0031] The specific embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit this disclosure.
[0032] It should be understood that the term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the following description.
[0033] It should be noted that the concepts of "first," "second," etc., mentioned in this disclosure are used only to distinguish different devices, modules, or units, and are not used to limit the order of functions performed by these devices, modules, or units or their interdependencies. The modifiers "a" and "a plurality of" mentioned in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless explicitly stated in the context, they should be understood as "one or more." In the description of this disclosure, unless otherwise stated, "a plurality of" means two or more, and other quantifiers are similar; "at least one," "one or more," or similar expressions refer to any combination of these items, including any combination of single or multiple items.
[0034] Although operations or steps are described in a specific order in the accompanying drawings in the embodiments of this disclosure, it should not be construed as requiring these operations or steps to be performed in the specific order or serial order shown, or requiring all of the shown operations or steps to be performed to obtain the desired result. In the embodiments of this disclosure, these operations or steps may be performed serially; they may be performed in parallel; or a portion of these operations or steps may be performed.
[0035] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of these messages or information. It is understood that before using the technical solutions disclosed in the embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.
[0036] Figure 1 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 1 As shown in the figure, this disclosure provides a method for issuing alarms on the quality of connected vehicle services, the method may include the following steps: In step S10, the historical network data of the vehicle network within a preset time range is preprocessed and aggregated according to multiple aggregation levels required by the business to obtain the dataset corresponding to each aggregation level. Different aggregation levels correspond to business datasets with different spatiotemporal dimensions.
[0037] In this step, historical network data of the vehicle network within a preset time range is preprocessed and aggregated according to multiple aggregation levels required by the business to obtain datasets corresponding to each aggregation level. Different aggregation levels correspond to business datasets with different spatiotemporal dimensions. For example, the historical network data can first be processed for missing and outlier values, then aggregated according to multiple aggregation levels required by the business to obtain candidate aggregated datasets. Finally, candidate aggregated datasets with a sample size below a set threshold are filtered out to obtain the datasets corresponding to each aggregation level.
[0038] In step S20, the datasets of the multiple aggregation levels are input into the hierarchical detection model for anomaly detection to obtain the anomaly detection results of each aggregation level dataset. Different aggregation levels correspond to different levels of detection sub-models in the hierarchical detection model.
[0039] In this step, datasets at multiple aggregation levels are input into a hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset. Different aggregation levels correspond to different levels of detection sub-models within the hierarchical detection model. For example, multiple aggregation levels may include first-level aggregation, second-level aggregation, third-level aggregation, and fourth-level aggregation. The first-level aggregation is used to aggregate vehicle-level spatiotemporal service datasets; the second-level aggregation is used to aggregate DNN (Data Network Name)-level spatiotemporal service datasets; the third-level aggregation is used to aggregate cell-level spatiotemporal service datasets; and the fourth-level aggregation is used to aggregate district / county-level spatiotemporal service datasets. The vehicle level is lower than the DNN level, the DNN level is lower than the cell level, and the cell level is lower than the district / county level. The hierarchical detection model may include first-level detection sub-models, second-level detection sub-models, third-level detection sub-models, and fourth-level detection sub-models corresponding to the first-level, second-level, third-level, and fourth-level aggregated datasets, respectively, used for anomaly detection on the aggregated datasets at their respective levels.
[0040] In step S30, based on the anomaly detection results of the datasets at each aggregation level, alarm suppression and integration are performed according to the priority of aggregation level from large to small, and alarm information is output.
[0041] In this step, based on the anomaly detection results of each aggregation level dataset, alarms are suppressed and integrated according to the priority from largest to smallest aggregation level, and alarm information is output. For example, according to the principle that high aggregation level anomaly alarms have higher priority than low aggregation level anomaly alarms, high aggregation level anomaly alarms are output first, the triggering of low aggregation level anomaly alarms is suppressed, and standardized alarm information corresponding to high aggregation level anomaly alarms is output.
[0042] This disclosure preprocesses historical network data of the vehicle-to-everything (V2X) network within a preset time range and aggregates it according to multiple aggregation levels required by the business to obtain datasets corresponding to each aggregation level. These datasets are then input into a hierarchical detection model for anomaly detection, yielding anomaly detection results for each aggregation level dataset. Based on the anomaly detection results of each aggregation level dataset, alarms are suppressed and integrated according to the priority from largest to smallest aggregation level, and alarm information is output. Because a hierarchical detection method is used for V2X data, it can significantly reduce missed and incorrect fault detections, improve detection accuracy, and enable the operation and maintenance team to effectively carry out network operation and maintenance work, improving fault resolution efficiency. The solution of this disclosure also has good detection performance for V2X data that integrates across different professions / businesses.
[0043] Figure 2 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 2 As shown, the preprocessing of historical network data of the vehicle network within a preset time range, and the aggregation according to multiple aggregation levels required by the business to obtain the dataset corresponding to each aggregation level, may include the following steps: In step S101, missing and outlier values are processed in the historical network data.
[0044] In this step, missing and outlier values are processed in historical network data to eliminate noise. Short-term fluctuations and small sample data are filtered to focus on long-term, large-scale anomalies; and data granularity is standardized by aggregating raw fine-grained data to the spatiotemporal dimensions required by the business (such as daily granularity or district / county scope); finally, data comparability is enhanced: standardization operations (such as Z-score) are used to eliminate dimensional differences between different indicators, facilitating data comparison. For example, outlier identification can be performed by configuring data value ranges (e.g., the range for successful paging should be ≥0). After processing outlier data, the accuracy of the aggregated data values can be avoided; for example, data that should be positive may become abnormal due to the inclusion of negative values.
[0045] In step S102, aggregation is performed according to multiple aggregation levels required by the business to obtain a candidate aggregation dataset.
[0046] In this step, aggregation is performed according to multiple aggregation levels required by the business to obtain candidate aggregated datasets. For example, vehicle network data can be aggregated to obtain a vehicle-level aggregated dataset, which may include spatiotemporal aggregation information and metric content. An example vehicle-level aggregated dataset is shown in Table 1. Table 1:
[0047] An example of hierarchical spatiotemporal aggregated information content can be shown in Table 2.
[0048] Table 2:
[0049] Then, based on the detection requirements, the index aggregation values are layered and aggregated. The aggregation rule is: the input index is calculated according to the corresponding spatiotemporal aggregation information to obtain the index aggregation value. The index aggregation value can include the aggregation value of any index among the 35 indicators in Table 1.
[0050] In step S103, candidate aggregate datasets with a sample count lower than a set threshold are filtered out to obtain the datasets corresponding to each aggregation level.
[0051] In this step, candidate aggregation datasets with fewer than a set threshold of samples are filtered out to obtain datasets corresponding to each aggregation level. For example, sample datasets with small-scale, short-term fluctuations can be filtered. The filtering rules are configurable and can be set to: aggregation level (single vehicle level, community level, DNN level, district / county level) and number of vehicles (≥ set threshold). For example, at the DNN level, only aggregation results with ≥50 vehicles are retained. The filtered sample data are not subjected to subsequent standardization processing and anomaly detection to avoid small sample statistical bias or accidentally added vehicles.
[0052] This process purifies and standardizes the data, providing high-quality data support that is adaptable to different detection scenarios for hierarchical anomaly detection, and ensuring the stability and accuracy of subsequent detection models.
[0053] In one possible implementation, feature standardization can be performed on the indicator features in the dataset. All aggregated indicators are considered as features, and the features are standardized by calculating the Z-score = (original value of an indicator - mean of an indicator) / standard deviation of an indicator. For example, to calculate the Z-score of the paging success rate at the DNN level, the current value, historical mean, and historical standard deviation (e.g., the past 30 days) of the paging success rate indicator can be taken. (99.3 - 98.8) / 3 ≈ 0.1667. After all indicators are calculated according to the above process, a dimensionless sample set is output. All indicators can be compared on the same scale, avoiding the problem of excessive weighting of individual indicators when calculating Mahalanobis distance, which affects the distance estimation.
[0054] Figure 3 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 3 As shown, the hierarchical detection model includes a first-level anomaly detection sub-model corresponding to the first-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain the anomaly detection results for each aggregation level dataset may include the following steps: In step S201a, the baseline mean and first covariance matrix of each indicator of the vehicle-level spatiotemporal business dataset are calculated using the minimum covariance determinant algorithm through the first-level anomaly detection sub-model.
[0055] In this step, the baseline mean μ1 and the first covariance matrix Σ1 of each indicator in the vehicle-level spatiotemporal service dataset are calculated using the minimum covariance determinant algorithm through the first-level anomaly detection sub-model. For example, calculating the baseline mean μ1 and the first covariance matrix Σ1 of each indicator collected at a time granularity (15 minutes / day) in the vehicle-level spatiotemporal service dataset using the minimum covariance determinant algorithm is existing technology and will not be elaborated further here. This algorithm can reduce the impact of outliers on baseline estimation and has good robustness, while also considering the correlation of multiple network indicators (such as high latency possibly accompanied by high packet loss rate), avoiding the one-sidedness of single-indicator detection.
[0056] In step S202a, based on the baseline mean and the first covariance matrix, the first Mahalanobis distance between each index vector of each vehicle and the corresponding baseline mean is calculated for each vehicle.
[0057] In this step, based on the baseline mean μ1 and the first covariance matrix Σ1, the first Mahalanobis distance D1 between each vehicle's index vector and the corresponding baseline mean is calculated. For example, the square of the first Mahalanobis distance D1 can be obtained by the following formula: D1 2 =( X μ1 ) T Σ1 1 ( X μ1 )Formula 1 in, T For the transpose operation, Σ1 1 Let Σ1 be the inverse of the first covariance matrix. X Standardize vectors for vehicle-level metrics (e.g., Z-score values that may include paging success rate, initial registration success rate, and PDU session establishment success rate).
[0058] In step S203a, if the first Mahalanobis distance is greater than the first preset threshold, the vehicle is determined to be abnormal.
[0059] In this step, if the first Mahalanobis distance D1 is greater than a first preset threshold, the vehicle is determined to be abnormal. The first preset threshold can be a dynamic threshold; for example, the square of the Mahalanobis distance follows a chi-square (...). 2 The distribution is chi-squared, so the first preset threshold is set using a chi-square distribution, and the degrees of freedom are set according to the number of indicators. And usually a 95% confidence level is used. Calculate the first preset threshold: .
[0060] In step S204a, if the vehicle is abnormal and there is no higher-level abnormality, a first list of abnormality indicators is output.
[0061] In this step, if a vehicle is abnormal and there is no higher-level (e.g., DNN level, community level, or district / county level) abnormality, a first list of abnormal indicators is output. For example, the first list of abnormal indicators may include the vehicle card number, the district / county to which it belongs, the granularity (e.g., 15 minutes / day), the time (e.g., 24 hours), indicator details (including the abnormal indicator name, indicator value, indicator threshold, deviation value (Mahavir distance)), and whether it is abnormal.
[0062] This process focuses on individual vehicle anomalies, fully considers the correlation of multiple indicators, improves the robustness and accuracy of vehicle anomaly identification, and makes it easier for maintenance personnel to quickly locate network problems in individual vehicles.
[0063] Figure 4 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 4 As shown, the hierarchical detection model includes a second-level anomaly detection sub-model corresponding to the second-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain the anomaly detection results for each aggregation level dataset may include the following steps: In step S201b, the second-level anomaly detection sub-model calculates the baseline values of each indicator and the second covariance matrix of the spatiotemporal business dataset at the DNN level for a specified region within the same time period in the past specific time window, based on the weighted moving average.
[0064] In this step, the second-level anomaly detection sub-model calculates the baseline values μ2 and the second covariance matrix Σ2 of each indicator in the spatiotemporal business dataset at the DNN level for a specified region (e.g., district / county) within the same time period (e.g., 08:00-09:00) over a specific time window (e.g., 7 days) in the specified DNN region (e.g., 08:00-09:00). For example, the baseline value μ2 can be obtained using the following formula: μ2= / Formula 2 in, For weight values, It was 7 days ago. The weighting is based on the most recent day, with the weighting increasing over the last 7 days (the most recent date has the highest weight). Standardize the vector values of the DNN-level index. Not zero.
[0065] The second covariance matrix Σ2 can be obtained by the following formula: Σ2 Formula 3 in, Not zero.
[0066] In step S202b, based on the baseline values of each indicator and the second covariance matrix, the deviation between each indicator vector of the spatiotemporal business dataset at the DNN level and the corresponding baseline value is calculated for the current time period. If the deviation is greater than a second preset threshold, the specified DNN is determined to be abnormal in the current time period.
[0067] In this step, based on the baseline values of each indicator and the second covariance matrix, the deviation of each indicator vector in the spatiotemporal business dataset at the DNN level for the current time period (e.g., the current hour) from its corresponding baseline value is calculated. If the deviation exceeds a second preset threshold, the specified DNN is determined to be abnormal in the current time period. For example, based on the baseline values of each indicator and the second covariance matrix, the deviation can be obtained first... Statistics, for example, The statistic can be obtained by the following formula: Formula 4 in, The second covariance matrix The inverse matrix of , where y is the DNN-level multi-index standardized vector for the current hour.
[0068] because The statistic follows Hotelling's... For ease of calculation, the distribution is converted into the F-value of the F-distribution. For example, the F-value of the F-distribution can be obtained by the following formula: Formula 5 Where p represents the number of indicators (p=35) and n represents the number of historical samples (n>p, for example n=60). Not zero.
[0069] Then, according to the degrees of freedom (35, 60) and F value, the F distribution table is consulted to obtain the deviation value. If the deviation value is >0.01 (second preset threshold), it is judged as statistically significant abnormality, and thus the specified DNN is judged to be abnormal in the current time period.
[0070] In step S203b, if the specified DNN is abnormal and there are no higher-level abnormalities, a second list of abnormality indicators is output.
[0071] In this step, if a DNN anomaly is specified and there are no higher-level anomalies (such as community-level or district-level anomalies), a second list of anomaly indicators is output. For example, the second list of anomaly indicators may include: DNN, district / county, time (current hour), indicator details (including anomaly indicator name, indicator value, indicator threshold, deviation value), and whether it is anomaly.
[0072] This method can capture joint anomalies of multiple indicators over time and ignore short-term degradation of single indicators, making the detection results more stable and accurate.
[0073] Figure 5 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 5 As shown, the hierarchical detection model includes a second-level anomaly detection sub-model corresponding to the second-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain the anomaly detection results for each aggregation level dataset may include the following steps: In step S204b, the second Mahalanobis distance between the specified DNN and other DNNs within the specified region is calculated in the multi-index space of the spatiotemporal business dataset at the DNN level. A horizontal comparison between DNNs is then performed by comparing the second Mahalanobis distance with a third preset threshold determined based on the chi-square distribution to determine whether there are any spatially isolated DNN anomalies.
[0074] In this step, the second Mahalanobis distance between the specified DNN and other DNNs within the specified region is calculated in the multi-index space of the spatiotemporal service dataset at the DNN level. A horizontal comparison between DNNs is then performed by comparing the second Mahalanobis distance with a third preset threshold determined based on the chi-square distribution to determine if there are any spatially isolated DNN anomalies. For example, the second Mahalanobis distance can be obtained by referring to step S202a, using the baseline mean and covariance matrix of each index of different DNNs collected within the same time period in the DNN-level spatiotemporal service dataset.
[0075] In step S205b, if the second Mahalanobis distance is greater than the third preset threshold, the specified DNN is determined to be abnormal.
[0076] In this step, if the second Mahalanobis distance is greater than the third preset threshold, a specified DNN anomaly is identified. For example, referring to step S203a, the third preset threshold can be set using a chi-square distribution, with degrees of freedom and confidence levels set according to the number of indicators, and the third preset threshold can be calculated.
[0077] In step S206b, if the specified DNN is abnormal and there are no higher-level abnormalities, a third list of abnormality indicators is output.
[0078] In this step, if a DNN anomaly is specified and there are no higher-level anomalies (such as community-level or district-level anomalies), a third list of anomaly indicators is output. For example, the third list of anomaly indicators may include: DNN, district / county, date, indicator details (including anomaly indicator name, indicator value, indicator threshold, deviation value (Mahavira distance)), and whether it is anomaly.
[0079] This method can perform horizontal comparisons between DNNs. Through spatial horizontal comparisons, it can accurately identify problems such as sudden increases in business load and abnormal DNN services, reducing misjudgments from single-dimensional detection.
[0080] Figure 6 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 6As shown, the hierarchical detection model includes a third-level anomaly detection sub-model corresponding to the third-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain the anomaly detection results for each aggregation level dataset may include the following steps: In step S201c, the third-level anomaly detection sub-model uses an exponential smoothing model to predict the current theoretical values of each indicator in the spatiotemporal service dataset of the specified cell based on historical network indicator data, and calculates the prediction deviation.
[0081] In this step, the third-level anomaly detection sub-model uses an exponential smoothing model based on historical network indicator data to predict the current theoretical values of each indicator in the spatiotemporal service dataset at the cell level for a specified cell, and calculates the prediction bias. For example, it can detect collective anomalies in network indicators (such as paging success rate and TCP establishment rate) for all vehicles in a specified cell at the hourly granularity. This process identifies regional problems caused by base station failures, localized signal interference, etc., through time-series mutation detection and comparison with neighboring cells.
[0082] It can retrieve the past 30 days' data for a specified neighborhood (the same time period each day, e.g., 8:00-9:00). It uses exponential smoothing (ETS) to predict the theoretical values of all indicators for the day. And calculate the prediction bias for all indicators, including positive bias: and negative deviation: This is used for subsequent calculation of cumulative deviation, where, This represents the actual value of the indicator.
[0083] In step S202c, a cumulative sum algorithm is used to continuously monitor the prediction deviation of at least one target indicator. When the cumulative deviation exceeds a fourth preset threshold, the designated cell is determined to be a persistent anomaly in the time series.
[0084] In this step, a cumulative sum algorithm is used to continuously monitor the prediction deviation of at least one target indicator. When the cumulative deviation exceeds a fourth preset threshold, the specified cell is determined to be a persistent anomaly in the time series. For example, cumulative deviation anomaly detection can be performed on all indicators of the specified cell (e.g., using the CUSUM algorithm), defining the cumulative deviation of the target indicator (such as paging success rate). initial value Perform iterative calculations, for example, cumulative positive and negative deviations. (where + represents positive and - represents negative) can be obtained from the following formula: Formula 6 in, Take the positive or negative deviation value calculated in the first step.k This is the allowable fluctuation range (usually taken as 50% of the historical error standard deviation). To obtain the maximum value.
[0085] The positive and negative cumulative biases are calculated simultaneously using a two-sided CUSUM method. If the value exceeds a fourth preset threshold h, the specified cell is determined to be a persistent anomaly in the time series. For example, the fourth preset threshold h can be obtained by the following formula: h=5×σ _历史标准差 Formula 7 Where, σ _历史标准差 The standard deviation of the target indicator over a historical period is given.
[0086] In step S203c, if the specified cell is abnormal and there is no higher-level abnormality, a fourth list of abnormal indicators is output.
[0087] In this step, if a specified cell is abnormal and there are no higher-level (e.g., district / county-level) anomalies, a fourth list of anomaly indicators is output. For example, the fourth list of anomaly indicators may include: cell code, district / county, time, indicator name, deviation value, threshold, and whether it is abnormal.
[0088] This algorithm can capture persistent anomalies over time, ignore short-term random fluctuations, and adaptively adjust the threshold based on historical data to adapt to dynamic changes in the network.
[0089] Figure 7 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 7 As shown, the hierarchical detection model includes a third-level anomaly detection sub-model corresponding to the third-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain the anomaly detection results for each aggregation level dataset may include the following steps: In step S204c, the differences of specified indicators between the specified cell and multiple physically neighboring cells are calculated, and a t-test is used to determine whether the differences of the specified indicators are significant, so as to determine whether the specified cell is a spatially isolated anomaly.
[0090] In this step, the differences in a specified metric (e.g., paging success rate) between the designated cell and multiple (e.g., 5) physically neighboring cells are calculated, and a t-test is used to determine whether the differences in the specified metric are significant, thereby determining whether the designated cell is a spatially isolated anomaly. For example, the significance index t can be obtained by the following formula: t = (Specified cell index value - Mean of neighboring cells) / (Standard deviation of neighboring cells) ) Formula 8 The neighboring cell mean is the mean of the specified indicators of all cells, including the specified cell and its neighboring cells. The neighboring cell standard deviation is the standard deviation of the specified indicators of all cells, including the specified cell and its neighboring cells. The number of neighboring cells is not zero.
[0091] like If so, the specified cell is determined to be an isolated anomalous cell in space.
[0092] In step S205c, if the specified cell is abnormal and there is no higher-level abnormality, the fifth abnormality indicator list is output.
[0093] In this step, if a specified cell is abnormal and there are no higher-level (e.g., district / county-level) anomalies, a fifth list of anomaly indicators is output. For example, the fifth list of anomaly indicators may include: cell code, district / county, time, deviation value, mean value, and whether it is abnormal.
[0094] This process, which combines spatial comparison with neighboring cells, can effectively identify regional problems such as base station failures and local signal interference, while ignoring short-term random fluctuations, thus improving the accuracy of cell anomaly detection.
[0095] Figure 8 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 8 As shown, the hierarchical detection model includes a fourth-level anomaly detection sub-model corresponding to the fourth-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain the anomaly detection results for each aggregation level dataset may include the following steps: In step S201d, the fourth-level anomaly detection sub-model is used to statistically analyze the proportion of abnormal vehicles in each district and county area of the district and county-level spatiotemporal business dataset within the sliding time window.
[0096] In this step, the fourth-level anomaly detection sub-model statistically analyzes the proportion of abnormal vehicles in each district / county within the spatiotemporal business dataset at the district / county level within a sliding time window. For example, it can detect anomalies in vehicle network indicators occurring simultaneously in multiple districts / counties within a short period (e.g., 1-2 hours), identifying global problems caused by core network failures, large-scale network attacks, or regional natural disasters. This method rapidly locates large-scale anomalies through real-time sliding window statistics and spatial correlation analysis, while avoiding misjudging local problems as global failures.
[0097] Specifically, based on the first list of abnormal indicators output in step S204a, a sliding window (e.g., 2 hours) is used to calculate the mean and standard deviation of the proportion of abnormal vehicles in each district and county.
[0098] In step S202d, a dynamic threshold for the proportion of abnormal vehicles within the sliding time window is calculated. The dynamic threshold is determined based on the mean and standard deviation of the historical proportion of abnormal vehicles within the time window.
[0099] In this step, a dynamic threshold for the proportion of abnormal vehicles within a sliding time window is calculated. The dynamic threshold is determined based on the mean and standard deviation of the historical proportion of abnormal vehicles within the time window. For example, the dynamic threshold can be determined using the 3σ principle, i.e., dynamic threshold = mean of the proportion of abnormal vehicles in the window + 3 × standard deviation of the proportion of abnormal vehicles in the window.
[0100] In step S203d, if the proportion of abnormal vehicles in the current district / county exceeds the dynamic threshold, it is determined that a large-scale anomaly has occurred in the current district / county, and a sixth list of abnormal indicators is output.
[0101] In this step, if the proportion of abnormal vehicles in the current district / county exceeds a dynamic threshold, it is determined that a large-scale anomaly has occurred in the current district / county, and a sixth list of anomaly indicators is output. For example, the sixth list of anomaly indicators may include: district / county, time, total number of vehicles, number of abnormal vehicles, whether it is abnormal, and anomaly proportion. This algorithm adaptively adjusts the threshold based on historical data within a time window to adapt to day-night traffic fluctuations and reduce false alarms.
[0102] This process, through real-time sliding window statistics and spatial correlation analysis, can quickly locate global issues such as core network failures and large-scale network attacks, while avoiding misjudging local problems as global failures, thus improving the timeliness and accuracy of responses to large-scale anomalies.
[0103] Figure 9 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 9 As shown, the process of suppressing and integrating alarms based on the anomaly detection results of the datasets at each aggregation level, according to the priority from largest to smallest aggregation level, and outputting alarm information, may include the following steps: In step S301, according to the principle that the priority of high-aggregation-level abnormal alarms is higher than that of low-aggregation-level abnormal alarms, high-aggregation-level abnormal alarms are output first, the triggering of low-aggregation-level abnormal alarms is suppressed, and the standardized alarm information corresponding to the high-aggregation-level abnormal alarms is output.
[0104] In this step, following the principle that high-aggregation-level anomaly alarms have higher priority than low-aggregation-level anomaly alarms, high-aggregation-level anomaly alarms are output first, while the triggering of low-aggregation-level anomaly alarms is suppressed. Standardized alarm information corresponding to high-aggregation-level anomaly alarms is then output. For example, anomaly detection results corresponding to each aggregation level can be collected and sorted according to the aggregation level from highest to lowest priority: district / county level is higher than cell level, cell level is higher than DNN level, and DNN level is higher than vehicle level. High-aggregation-level anomaly alarms are triggered and output first, while the triggering of low-aggregation-level anomaly alarms is suppressed to avoid duplicate alarms caused by the same root cause. Finally, standardized alarm information corresponding to high-aggregation-level anomalies is output, including key information such as the anomaly object, time, and indicator details.
[0105] This process streamlines the number of alarms, clarifies operational priorities, helps operations personnel quickly focus on core issues, and significantly improves alarm handling efficiency.
[0106] Figure 10 This is a flowchart illustrating a method for issuing alarms for vehicle-to-everything (V2X) service quality according to an embodiment of this disclosure. Figure 10 As shown, the method may further include the following steps: In step S40, fault events confirmed by maintenance personnel are collected, and based on the fault events, the algorithm parameters involved in at least one anomaly detection sub-model in the hierarchical detection model are optimized.
[0107] In this step, fault events confirmed by operations and maintenance personnel are collected, and based on these events, the algorithm parameters of at least one anomaly detection sub-model in the hierarchical detection model are optimized. For example, actual fault events confirmed by operations and maintenance personnel can be continuously collected, and the corresponding network data and anomaly detection results can be organized and analyzed. Based on the feedback from these fault events, the algorithm parameters of each anomaly detection sub-model in the hierarchical detection model are adjusted accordingly, including baseline calculation parameters, various thresholds, and sliding window size. By continuously feeding real-world fault cases back into the model, dynamic optimization of the algorithm parameters is achieved, making the hierarchical detection model more closely aligned with actual network operation scenarios, continuously improving the accuracy and adaptability of anomaly detection, and extending the effective lifespan of the model.
[0108] Figure 11 This is a block diagram of a vehicle-to-everything (V2X) service quality alarm device according to an embodiment of this disclosure. Figure 11 As shown in the figure, this disclosure provides a vehicle networking service quality alarm device 1100, which may include the following modules: The aggregation module 1110 is used to preprocess the historical network data of the Internet of Vehicles within a preset time range, and aggregate it according to multiple aggregation levels required by the business to obtain the datasets corresponding to each aggregation level. Different aggregation levels correspond to business datasets with different spatiotemporal dimensions.
[0109] The detection module 1120 is used to input the datasets of the multiple aggregation levels into the hierarchical detection model for anomaly detection, so as to obtain the anomaly detection results of each aggregation level dataset. Different aggregation levels correspond to different levels of detection sub-models in the hierarchical detection model.
[0110] The output module 1130 is used to suppress and integrate alarms based on the anomaly detection results of the datasets at each aggregation level, according to the priority of the aggregation level from large to small, and output alarm information.
[0111] In one possible implementation, the aggregation module 1110 is further configured to: The historical network data is processed for missing and outlier values. Aggregate according to multiple aggregation levels required by the business to obtain candidate aggregation datasets; Candidate aggregated datasets with a sample size below a set threshold are filtered out to obtain the datasets corresponding to each aggregation level; The multiple aggregation levels include a first-level aggregation, a second-level aggregation, a third-level aggregation, and a fourth-level aggregation. The first-level aggregation is used to aggregate vehicle-level spatiotemporal service datasets; the second-level aggregation is used to aggregate DNN-level spatiotemporal service datasets; the third-level aggregation is used to aggregate cell-level spatiotemporal service datasets; and the fourth-level aggregation is used to aggregate district / county-level spatiotemporal service datasets. The vehicle level is lower than the DNN level, the DNN level is lower than the cell level, and the cell level is lower than the district / county level.
[0112] In one possible implementation, the detection module 1120 is further configured to: Using the first-level anomaly detection sub-model, the baseline mean and first covariance matrix of each indicator in the vehicle-level spatiotemporal business dataset are calculated using the minimum covariance determinant algorithm. Based on the baseline mean and the first covariance matrix, calculate the first Mahalanobis distance between each vehicle's index vector and the corresponding baseline mean; If the first Mahalanobis distance is greater than a first preset threshold, the vehicle is determined to be abnormal. If the vehicle is abnormal and there are no higher-level abnormalities, output the first list of abnormal indicators.
[0113] In one possible implementation, the detection module 1120 is further configured to: The second-level anomaly detection sub-model calculates the baseline values of each indicator and the second covariance matrix of the spatiotemporal business dataset of the specified region within the same time period in the past specific time window using the weighted moving average; Based on the baseline values of each indicator and the second covariance matrix, the deviation between each indicator vector of the spatiotemporal business dataset at the DNN level and the corresponding baseline value is calculated for the current time period, and if the deviation is greater than the second preset threshold, the specified DNN is determined to be abnormal in the current time period. If the specified DNN is anomaly, and there are no higher-level anomalies, output a second list of anomaly indicators; or, Calculate the second Mahalanobis distance between the specified DNN and other DNNs within the specified region in the multi-index space of the spatiotemporal business dataset at the DNN level, and perform a horizontal comparison between DNNs by comparing the second Mahalanobis distance with a third preset threshold determined based on the chi-square distribution, in order to determine whether there are any spatially isolated DNN anomalies. If the second Mahalanobis distance is greater than the third preset threshold, the specified DNN is determined to be abnormal; If the specified DNN is abnormal and there are no higher-level abnormalities, output a third list of abnormality indicators.
[0114] In one possible implementation, the detection module 1120 is further configured to: Using the third-level anomaly detection sub-model, based on historical network indicator data, the exponential smoothing model is used to predict the current theoretical values of each indicator in the spatiotemporal service data of the specified cell, and the prediction deviation is calculated. The prediction deviation of at least one target indicator is continuously monitored using a cumulative sum algorithm. When the cumulative deviation exceeds a fourth preset threshold, the designated cell is determined to be a persistent anomaly in the time series. If the specified cell is abnormal, and there are no higher-level abnormalities, output a fourth list of abnormal indicators; or, Calculate the differences of specified indicators between the specified cell and multiple physically neighboring cells, and use a t-test to determine whether the differences of the specified indicators are significant, so as to determine whether the specified cell is a spatially isolated anomaly. If the specified cell is abnormal and there are no higher-level abnormalities, output the fifth list of abnormal indicators.
[0115] In one possible implementation, the detection module 1120 is further configured to: Using the fourth-level anomaly detection sub-model, the proportion of abnormal vehicles in each district and county area of the district and county-level spatiotemporal business dataset is statistically analyzed within the sliding time window. Calculate the dynamic threshold of the proportion of abnormal vehicles within the sliding time window, the dynamic threshold being determined based on the mean and standard deviation of the historical proportion of abnormal vehicles within the time window; If the proportion of abnormal vehicles in the current district or county exceeds the dynamic threshold, it is determined that a large-scale anomaly has occurred in the current district or county, and a sixth list of abnormal indicators is output.
[0116] In one possible implementation, the output module 1130 is further configured to: Following the principle that high-aggregation-level anomaly alarms have higher priority than low-aggregation-level anomaly alarms, high-aggregation-level anomaly alarms are output first, the triggering of low-aggregation-level anomaly alarms is suppressed, and the standardized alarm information corresponding to the high-aggregation-level anomaly alarms is output.
[0117] In one possible implementation, the device 1100 further includes a feedback module for: Collect fault events confirmed by operation and maintenance personnel, and based on the fault events, optimize the algorithm parameters involved in at least one anomaly detection sub-model in the hierarchical detection model.
[0118] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.
[0119] In summary, the aggregation module in the device is responsible for collecting historical network data of the vehicle-to-everything (V2X) network within a preset time range, handling missing and outlier values, aggregating data according to multiple aggregation levels, filtering out unqualified candidate datasets, and outputting the valid datasets corresponding to each aggregation level. The detection module inputs the datasets of each aggregation level into the hierarchical detection model, and performs anomaly detection through different levels of detection sub-models, generating corresponding anomaly detection results. The output module performs alarm suppression and integration on the detection results according to the priority of aggregation level from largest to smallest, and outputs standardized alarm information. Through efficient collaboration between modules, this device achieves systematic and accurate detection and alarming of V2X service quality anomalies, providing strong support for operation and maintenance work.
[0120] Figure 12 This is a block diagram illustrating an electronic device according to an exemplary embodiment. Figure 12 As shown, the electronic device 1200 may include: a processor 1201 and a memory 1202. The electronic device 1200 may also include one or more of a multimedia component 1203, an input / output (I / O) interface 1204, and a communication component 1205.
[0121] The processor 1201 controls the overall operation of the electronic device 1200 to complete all or part of the steps in the aforementioned vehicle network service quality alarm method. The memory 1202 stores various types of data to support the operation of the electronic device 1200. This data may include, for example, instructions for any application or method operating on the electronic device 1200, and application-related data such as contact data, sent and received messages, pictures, audio, video, etc. The memory 1202 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. Multimedia component 1203 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in memory 1202 or transmitted via communication component 1205. The audio component also includes at least one speaker for outputting audio signals. I / O interface 1204 provides an interface between processor 1201 and other interface modules, such as a keyboard, mouse, buttons, etc. These buttons may be virtual or physical buttons. Communication component 1205 is used for wired or wireless communication between the electronic device 1200 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IoT, eMTC, or other 5G technologies, or combinations thereof, is not limited here. Therefore, the corresponding communication component 1205 may include: a Wi-Fi module, a Bluetooth module, an NFC module, etc.
[0122] In an exemplary embodiment, the electronic device 1200 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described vehicle-to-everything (V2X) service quality alarm method.
[0123] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, which, when executed by a processor, implements the steps of the above-described vehicle-to-everything (V2X) service quality alarm method. For example, the computer-readable storage medium may be the memory 1202 including program instructions, which may be executed by the processor 1201 of the electronic device 1200 to complete the above-described V2X service quality alarm method.
[0124] In another exemplary embodiment, a computer program product is also provided, the computer program product comprising a computer program executable by a programmable device, the computer program having a code portion for performing the above-described vehicle network service quality alarm method when executed by the programmable device.
[0125] The preferred embodiments of this disclosure have been described in detail above with reference to the accompanying drawings. However, this disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this disclosure, various simple modifications can be made to the technical solutions of this disclosure, and these simple modifications all fall within the protection scope of this disclosure.
[0126] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, this disclosure will not describe the various possible combinations separately.
[0127] Furthermore, various different embodiments of this disclosure can be combined in any way, as long as they do not violate the spirit of this disclosure, they should also be regarded as the content disclosed in this disclosure.
Claims
1. A method for issuing alarms regarding the quality of connected vehicle services, characterized in that, The method includes: Historical network data of the Internet of Vehicles within a preset time range is preprocessed and aggregated according to multiple aggregation levels required by the business to obtain the datasets corresponding to each aggregation level. Different aggregation levels correspond to business datasets with different spatiotemporal dimensions. The datasets at the multiple aggregation levels are input into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset. Different aggregation levels correspond to different levels of detection sub-models in the hierarchical detection model. Based on the anomaly detection results of the datasets at each aggregation level, alarms are suppressed and integrated according to the priority of aggregation level from large to small, and alarm information is output.
2. The method according to claim 1, characterized in that, The process of preprocessing historical network data of the vehicle network within a preset time range and aggregating it according to multiple aggregation levels required by the business to obtain datasets corresponding to each aggregation level includes: The historical network data is processed for missing and outlier values. Aggregate according to multiple aggregation levels required by the business to obtain candidate aggregation datasets; Candidate aggregated datasets with a sample size below a set threshold are filtered out to obtain the datasets corresponding to each aggregation level; The multiple aggregation levels include a first-level aggregation, a second-level aggregation, a third-level aggregation, and a fourth-level aggregation. The first-level aggregation is used to aggregate spatiotemporal service datasets at the vehicle level; the second-level aggregation is used to aggregate spatiotemporal service datasets at the data network name (DNN) level; the third-level aggregation is used to aggregate spatiotemporal service datasets at the cell level; and the fourth-level aggregation is used to aggregate spatiotemporal service datasets at the district / county level. The vehicle level is lower than the DNN level, the DNN level is lower than the cell level, and the cell level is lower than the district / county level.
3. The method according to claim 2, characterized in that, The hierarchical detection model includes a first-level anomaly detection sub-model corresponding to the first-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset includes: Using the first-level anomaly detection sub-model, the baseline mean and first covariance matrix of each indicator in the vehicle-level spatiotemporal business dataset are calculated using the minimum covariance determinant algorithm. Based on the baseline mean and the first covariance matrix, calculate the first Mahalanobis distance between each vehicle's index vector and the corresponding baseline mean; If the first Mahalanobis distance is greater than a first preset threshold, the vehicle is determined to be abnormal. If the vehicle is abnormal and there are no higher-level abnormalities, output the first list of abnormal indicators.
4. The method according to claim 2, characterized in that, The hierarchical detection model includes a second-level anomaly detection sub-model corresponding to the second-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset includes: The second-level anomaly detection sub-model calculates the baseline values of each indicator and the second covariance matrix of the spatiotemporal business dataset of the specified region within the same time period in the past specific time window using the weighted moving average; Based on the baseline values of each indicator and the second covariance matrix, the deviation between each indicator vector of the spatiotemporal business dataset at the DNN level and the corresponding baseline value is calculated for the current time period, and if the deviation is greater than the second preset threshold, the specified DNN is determined to be abnormal in the current time period. If the specified DNN is anomaly, and there are no higher-level anomalies, output a second list of anomaly indicators; or, Calculate the second Mahalanobis distance between the specified DNN and other DNNs within the specified region in the multi-index space of the spatiotemporal business dataset at the DNN level, and perform a horizontal comparison between DNNs by comparing the second Mahalanobis distance with a third preset threshold determined based on the chi-square distribution, in order to determine whether there are any spatially isolated DNN anomalies. If the second Mahalanobis distance is greater than the third preset threshold, the specified DNN is determined to be abnormal; If the specified DNN is abnormal and there are no higher-level abnormalities, output a third list of abnormality indicators.
5. The method according to claim 2, characterized in that, The hierarchical detection model includes a third-level anomaly detection sub-model corresponding to the third-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset includes: Using the third-level anomaly detection sub-model, based on historical network indicator data, the current theoretical values of each indicator in the spatiotemporal service data of the specified cell are predicted using an exponential smoothing model, and the prediction deviation is calculated. The prediction deviation of at least one target indicator is continuously monitored using a cumulative sum algorithm. When the cumulative deviation exceeds a fourth preset threshold, the designated cell is determined to be a persistent anomaly in the time series. If the specified cell is abnormal, and there are no higher-level abnormalities, output a fourth list of abnormal indicators; or, Calculate the differences of specified indicators between the specified cell and multiple physically neighboring cells, and use a t-test to determine whether the differences of the specified indicators are significant, so as to determine whether the specified cell is a spatially isolated anomaly. If the specified cell is abnormal and there are no higher-level abnormalities, output the fifth list of abnormal indicators.
6. The method according to claim 2, characterized in that, The hierarchical detection model includes a fourth-level anomaly detection sub-model corresponding to the fourth-level aggregation. The step of inputting the datasets from the multiple aggregation levels into the hierarchical detection model for anomaly detection to obtain anomaly detection results for each aggregation level dataset includes: Using the fourth-level anomaly detection sub-model, the proportion of abnormal vehicles in each district and county area of the district and county-level spatiotemporal business dataset is statistically analyzed within the sliding time window. Calculate the dynamic threshold of the proportion of abnormal vehicles within the sliding time window, the dynamic threshold being determined based on the mean and standard deviation of the historical proportion of abnormal vehicles within the time window; If the proportion of abnormal vehicles in the current district or county exceeds the dynamic threshold, it is determined that a large-scale anomaly has occurred in the current district or county, and a sixth list of abnormal indicators is output.
7. The method according to claim 1, characterized in that, The method further includes: Collect fault events confirmed by operation and maintenance personnel, and based on the fault events, optimize the algorithm parameters involved in at least one anomaly detection sub-model in the hierarchical detection model.
8. An electronic device, characterized in that, include: A memory on which computer programs are stored; A processor for executing the computer program in the memory to implement the steps of the method according to any one of claims 1-7.
9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method described in any one of claims 1-7.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-7.