Digital television data analysis method and system based on multi-source data
Patent Information
- Application Number
- CN202610739802.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-27
- Publication Date
- 2026-10-23
AI Technical Summary
这导致运维人员面临大量无差别告警，难以快速定位最关键的中断节目，延误了恢复时机，使得运维效率低下，用户体验受损
[0012]本申请在故障发生后，快速查找所有与故障通道存在柔性关联的内容，生成受影响内容清单，确保了故障影响范围的完整识别。分析业务权重和冗余覆盖率缺口，能够量化每个内容面临的恢复紧迫程度。计算抢修紧迫度指数总和，将多个受影响内容的紧急程度聚合为一个整体指标，便于运维人员判断故障的全局严重性。为后续分级操作提供了明确的数值依据，实现了从“定性判断”到“定量决策”的升级。
Smart Images
Figure CN122970243A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of television data analysis technology, specifically to a digital television data analysis method and system based on multi-source data. Background Technology
[0002] When a transmission channel fails, existing digital television systems typically switch to a pre-defined, fixed backup channel. However, if a channel is currently in good health, but its health score has been declining over the past few days, it indicates potential deterioration and a higher risk of failure. Existing systems fail to recognize this trend and continue to use it as a reliable backup resource, increasing the risk of secondary failures. Regarding the urgency assessment of fault response, existing technologies often adopt a "one-size-fits-all" approach, applying the same alarm and handling procedures regardless of the importance or urgency of the affected programs. This results in maintenance personnel facing a large number of indiscriminate alarms, making it difficult to quickly locate the most critical interrupted programs, delaying recovery time, leading to low maintenance efficiency and a degraded user experience.
[0003] Therefore, this invention discloses a digital television data analysis method and system based on multi-source data to solve the above problems. Summary of the Invention
[0004] The purpose of this invention is to provide a digital television data analysis method and system based on multi-source data to solve the problems raised in the prior art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a digital television data analysis method based on multi-source data, the method comprising the following steps: S1: Collect audiovisual content data and transmission channel data, analyze the feature sequences of audiovisual content data and transmission channel data; analyze the correlation based on the matching degree of feature sequences; S2: Statistically analyze the abnormal time overlap of audiovisual content data and transmission channel data in historical faults; combine the co-occurrence frequency and matching degree of the attributes of audiovisual content data and transmission channel data to analyze the dependency of audiovisual content data and transmission channel data. S3: Based on the dependency between audiovisual content data and transmission channel data, generate a set of candidate transmission channels for audiovisual content data; analyze the availability confidence of candidate transmission channels when the primary transmission channel fails, and filter to generate an effective redundancy set and a failed redundancy set; S4: Generate a list of affected content and the service weight of each audiovisual content data based on the faulty transmission channel; integrate fault data to analyze the urgency index of emergency repair, and generate graded alarm signals and switching operation sequences based on the emergency repair urgency threshold.
[0006] S1 includes the following: Acquire audiovisual content data and transmission channel data; For the audiovisual content data, a frame is extracted every few seconds, and each extracted frame is scaled to a grayscale image of a predetermined size of pixels. The sign of the difference between the grayscale value of each pixel and the average grayscale value of the entire image is calculated. A binary sequence is generated based on the sign of the difference between all pixels as the visual fingerprint at the corresponding moment. Several visual fingerprints are continuously collected to form a content fingerprint sequence. For the data in the transmission channel, a transmission packet is captured every few seconds. The instantaneous bit rate, clock reference jitter value and retransmission request count are extracted from each transmission packet. The three indicators are normalized and combined into a three-dimensional vector as the time delay fingerprint at the corresponding moment. Several time delay fingerprints are continuously collected to form a channel fingerprint sequence. The position of the content fingerprint sequence is fixed, and the channel fingerprint sequence is slid forward and backward on the time axis respectively, with the total sliding length being a preset constant; for each offset, the fingerprint similarity of the two sequences at each corresponding position in the overlapping time period after the offset is calculated, and the average value is taken as the matching degree corresponding to the offset; the fingerprint similarity is obtained through cosine similarity analysis of the two sequences; Take the maximum matching degree under all offsets as the best matching degree, and record the optimal offset corresponding to the best matching degree; Obtain the load rate of the current transmission channel, which is the ratio of the current throughput to the rated capacity; denot the sum of the basic association threshold and the modified association threshold as the dynamic association threshold, which is equal to the product of the basic association threshold and the load rate; If the optimal matching degree is greater than the dynamic association threshold, it is determined that there is a flexible association relationship between the audiovisual content data and the transmission channel data, and the optimal offset and optimal matching degree corresponding to the flexible association relationship are recorded. For the same audiovisual content data, all flexible association relationships are aggregated to generate an association tuple; the association tuple includes: audiovisual content data, the content fingerprint sequence corresponding to the audiovisual content data, all transmission channels associated with the audiovisual content data, the corresponding optimal offset set, and the corresponding optimal matching degree set. This application determines the association between audiovisual content and transmission channels by calculating the matching degree between content fingerprint sequences and channel fingerprint sequences, avoiding the inefficiency and inaccuracy of traditional static binding or manual configuration. Matching degree analysis quantifies the degree of fit between content and channels, providing a reliable numerical basis for subsequent dependency assessment and redundancy screening. Compared to fixed binding, this application can adapt to dynamic changes in content and channels, improving the real-time performance and robustness of the association. The content fingerprint uses a scaled grayscale image + difference sign to generate a binary sequence, requiring no complex neural network and having low computational cost, making it suitable for real-time processing; at the same time, slight brightness changes do not affect the difference sign, providing anti-interference capabilities. The channel fingerprint uses three key indicators: instantaneous bit rate, clock reference jitter, and retransmission request count, comprehensively reflecting the core dimensions of transmission quality, and normalization facilitates similarity calculation. Continuously collecting multiple fingerprints to form a sequence is more stable than single-frame matching, eliminating instantaneous noise interference and improving matching accuracy.
[0007] S2 includes the following: For any associated tuple, extract the occurrence time of all content abnormal events from the audiovisual content data of the associated tuple from the historical fault log. The content abnormal events include, but are not limited to, screen freeze, audio-visual desynchronization, and mosaic burst. Extract the occurrence time of all channel abnormal events from the transmission channel data of the associated tuple from the historical fault log. The channel abnormal events include, but are not limited to, bit rate drop, clock reference jump, and retransmission request count surge. For each content abnormal event, if there is at least one channel abnormal event within a time window of several seconds before and after the occurrence time, then the content abnormal event and the channel abnormal event are counted as one associated abnormality. Count the number of associated anomalies and the total number of channel anomalies in the associated tuples; if the total number of channel anomalies is greater than zero, the ratio of the number of associated anomalies to the total number of channel anomalies is recorded as the fault propagation probability; if the total number of channel anomalies is equal to zero, the fault propagation probability is zero. Extract the static attributes of each audiovisual content data and each transmission channel data separately; The static attributes of audiovisual content data include bitrate level, screen size, and content category; the bitrate level includes, but is not limited to, low, medium, and high; the screen size includes, but is not limited to, standard definition, high definition, and ultra-high definition; the content category includes, but is not limited to, news, sports events, dramas, and variety shows; Static attributes of transmission channel data include, but are not limited to, bandwidth level, carrier, and coverage area; A co-occurrence count matrix is constructed based on the static attributes of audiovisual content data and transmission channel data. The row index of the co-occurrence count matrix corresponds to each specific attribute in the static attribute category of the audiovisual content data, and the column index corresponds to each specific attribute in the static attribute category of the transmission channel data. The matrix elements represent the number of times the static attributes of the audiovisual content data and the static attributes of the transmission channel data co-occur in all historical flexible association relationships. For the current content channel pair, the co-occurrence count under each attribute dimension combination is queried, and the co-occurrence counts corresponding to each attribute dimension combination are normalized and then weighted and summed, which is recorded as the comprehensive co-occurrence dependency degree. The normalized fault propagation probability, the comprehensive co-occurrence dependency, and the best matching degree are weighted and summed to form the comprehensive dependency coefficient. The transmission channel with the largest comprehensive dependency coefficient is used as the primary transmission channel for audiovisual content data.
[0008] This application quantitatively characterizes the transmission risk of channel failures to content quality by statistically analyzing the temporal overlap between content anomalies and channel anomalies in historical failure logs, reflecting a causal relationship rather than a simple correlation. Using a time window to determine associated anomalies allows for tolerance of certain transmission delays and timing errors, improving statistical accuracy. The calculated failure transmission probability directly reflects the content's sensitivity to channel failures, providing crucial historical evidence for dependency assessment. By constructing a co-occurrence counting matrix, the historical co-occurrence patterns of static attributes of content and channels are digitized, leveraging big data statistical patterns. Statistical analysis and weighted summation of multiple attribute dimensions separately capture the varying importance between different attribute pairs; for example, the co-occurrence of bitrate and bandwidth is more critical than the co-occurrence of category and operator, improving the granularity of dependency assessment.
[0009] S3 includes the following: For the audiovisual content data of the determined primary transmission channel, from the other transmission channels besides the primary transmission channel, sort them from high to low according to the comprehensive dependency coefficient, select the top few transmission channels, and generate a set of candidate transmission channels for the audiovisual content data. Obtain the health score sequence of each candidate transmission channel for the most recent few days, with the number of days as the independent variable and the health score as the dependent variable, and use linear regression to fit the score decay line of each candidate transmission channel to generate the slope of the score decay line. Obtain the average reserved bandwidth of the candidate transmission channel within the historical time window that is occupied by other audiovisual content data or tasks; analyze the remaining available bandwidth of the candidate transmission channel; the remaining available bandwidth is equal to the rated bandwidth of the candidate transmission channel minus the current real-time occupied bandwidth minus the average reserved bandwidth. Analyze the availability confidence of the backup transmission channel when the primary transmission channel fails; For each candidate transmission channel, if the availability confidence is greater than or equal to the availability confidence threshold, and the candidate transmission channel is not currently reserved exclusively by audiovisual content data with higher weight, then the candidate transmission channel is included in the effective redundancy set; the fact that the candidate transmission channel is not currently reserved exclusively by audiovisual content data with higher weight means that in the resource reservation snapshot of the candidate transmission channel, there is no exclusive reservation from other audiovisual content data with higher priority than the current audiovisual content data; the resource reservation snapshot is used to record the total amount of reserved bandwidth for each candidate transmission channel within a future time window that has been reserved by other audiovisual content data or tasks; Otherwise, the candidate transmission channels will be included in the set of failed redundancies.
[0010] This application selects candidate channels based on a comprehensive dependency coefficient ranking, ensuring a good fit between the candidate channels and the content, and avoiding blind selection. Constructing a decay line and calculating its slope based on the health score sequence can predict the health trend of channels, providing more foresight than simply using the current health value. Introducing reserved bandwidth average analysis considers resource competition among multiple programs, avoiding the risk of candidate channels being unusable due to bandwidth occupation in the event of a failure. The availability confidence score integrates remaining available bandwidth and health trends, and sets confidence thresholds and exclusive reservation checks, resulting in a more effective redundant set with higher actual availability, reducing the probability of failover failures.
[0011] S4 includes the following: Obtain the faulty transmission channel and the time of the fault; find all related tuples that have flexible relationships with the faulty channel, extract the audio-visual content data in each related tuple, and generate a list of affected content; analyze the business weight of each audio-visual content data in the list of affected content. Analyze the redundancy coverage gap of each audiovisual content data in the affected content list; the redundancy coverage gap is equal to the ratio of the number of candidate transmission channels in the failed redundancy set to the number of candidate transmission channels in the candidate transmission channel set. The repair urgency index is the sum of the repair urgency indices of the audiovisual content data in the affected content list; the repair urgency index is equal to the product of the business weight of the audiovisual content data, the redundancy coverage gap, and the interruption duration amplification factor; the interruption duration amplification factor is equal to the ratio of the duration from the time of the fault to the current time to the preset duration factor. If the calculated interrupt duration amplification factor is greater than the interrupt duration amplification factor threshold, then the corresponding interrupt duration amplification factor is fixed at the interrupt duration amplification factor threshold. Based on the sum of the emergency repair urgency indices, a tiered operation will be performed: If the total emergency repair urgency index is less than the first threshold for emergency repair urgency, a blue alert level will be triggered, which will only record the event without triggering any action. If the sum of the emergency repair urgency indices is greater than or equal to the first emergency repair urgency threshold but less than the second emergency repair urgency threshold, a yellow alarm level will be triggered. Select the candidate transmission channel with the highest available confidence from the corresponding effective redundancy set and match it with the corresponding audio-visual content data to generate a one-click switching operation sequence. The audio-visual content data in the one-click switching operation sequence are arranged in descending order from high to low according to the product of business weight and redundancy coverage gap, and pushed to the operation and maintenance console. If the repair urgency index is greater than or equal to the second repair urgency threshold, a red alarm level is triggered, along with an audible and visual alarm. A heatmap is displayed on the large screen showing the product of the redundancy coverage gap and the business weight of each audiovisual content data. A sequence of switching operations is output, arranged in descending order from high to low according to the product of the business weight and the redundancy coverage gap of the audiovisual content data. For audiovisual content data with an empty effective redundancy set, the output switching operation instruction includes a suggestion to reduce the bitrate or switch to backup content.
[0012] This application rapidly identifies all content with flexible connections to the faulty channel after a failure occurs, generating a list of affected content to ensure complete identification of the failure's impact scope. Analyzing business weights and redundancy coverage gaps quantifies the urgency of recovery for each piece of content. Calculating the sum of repair urgency indices aggregates the urgency of multiple affected contents into a single overall indicator, facilitating operations personnel's assessment of the overall severity of the failure. This provides clear numerical support for subsequent tiered operations, achieving an upgrade from "qualitative judgment" to "quantitative decision-making."
[0013] A digital television data analysis system based on multi-source data is implemented using the aforementioned digital television data analysis method based on multi-source data. The system includes a correlation analysis module, a transmission dependency analysis module, a candidate screening module, and a rescue index analysis module. The correlation analysis module is used to collect audiovisual content data and transmission channel data, analyze the feature sequences of audiovisual content data and transmission channel data, and analyze correlation relationships based on the matching degree of feature sequences. The transmission dependency analysis module is used to statistically analyze the abnormal time overlap between audiovisual content data and transmission channel data in historical faults; and to analyze the dependency between audiovisual content data and transmission channel data by combining the co-occurrence frequency and matching degree of the attributes of audiovisual content data and transmission channel data. The candidate screening module is used to generate a set of candidate transmission channels for audiovisual content data based on the dependency between audiovisual content data and transmission channel data; analyze the availability confidence of candidate transmission channels when the primary transmission channel fails, and filter to generate an effective redundancy set and a failed redundancy set. The emergency response index analysis module is used to generate a list of affected content and the business weight of each audiovisual content data based on the faulty transmission channel; it integrates fault data to analyze the emergency repair urgency index and generates graded alarm signals and switching operation sequences based on the emergency repair urgency threshold.
[0014] The association analysis module includes a fingerprint feature analysis unit and an association matching unit; The fingerprint feature analysis unit is used to acquire audiovisual content data and transmission channel data; and to generate content fingerprint sequences and channel fingerprint sequences based on the audiovisual content data and transmission channel data, respectively. The association matching unit is used to analyze the matching degree between the content fingerprint sequence and the channel fingerprint sequence, and determine the association relationship between the audiovisual content data and the transmission channel data based on the matching degree. The transmission dependency analysis module includes a dependency feature analysis unit and a primary transmission analysis unit; The dependency feature analysis unit is used to analyze the fault propagation probability and overall co-occurrence dependency of audiovisual content data and data of each transmission channel; The primary transmission analysis unit is used to weight and sum the normalized fault propagation probability, comprehensive co-occurrence dependency degree and best matching degree, which is denoted as the comprehensive dependency coefficient. The transmission channel with the largest comprehensive dependency coefficient is used as the primary transmission channel for audiovisual content data. The candidate screening module includes a candidate screening unit, an available confidence analysis unit, and an available confidence screening unit; The candidate screening unit is used to select the top several transmission channels from the remaining transmission channels other than the primary transmission channel, sorted by comprehensive dependency coefficient from high to low, to generate a candidate transmission channel set for the audiovisual content data of the audiovisual content data. The available confidence analysis unit is used to construct a score decay line based on the health score sequence of the candidate transmission channel and generate the slope of the score decay line; obtain the average amount of reserved bandwidth of the candidate transmission channel for other audiovisual content data or tasks within the historical time window; analyze the remaining available bandwidth of the candidate transmission channel; and analyze the availability confidence of the candidate transmission channel when the primary transmission channel fails based on the remaining available bandwidth and the slope of the score decay line. The available confidence screening unit is used to classify each candidate transmission channel into the effective redundancy set if the available confidence is greater than or equal to the available confidence threshold and the candidate transmission channel is not currently reserved exclusively by audiovisual content data with higher weight; otherwise, the candidate transmission channel is classified into the failed redundancy set.
[0015] The emergency rescue index analysis module includes an emergency repair characteristic analysis unit, an emergency repair urgency analysis unit, and an early warning unit. The emergency repair feature analysis unit is used to obtain the fault transmission channel and the time of the fault occurrence; find all related tuples that have flexible correlation with the fault channel, extract the audio-visual content data in each related tuple, and generate a list of affected content; analyze the service weight of each audio-visual content data in the list of affected content; and analyze the redundancy coverage gap of each audio-visual content data in the list of affected content. The emergency repair urgency analysis unit analyzes the sum of the emergency repair urgency indices of the audiovisual content data in the list of affected content; The early warning unit is used to perform graded operations based on the sum of the emergency repair urgency indices.
[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: This application determines the association between audiovisual content and transmission channels by calculating the matching degree between content fingerprint sequences and channel fingerprint sequences, avoiding the inefficiency and inaccuracy of traditional static binding or manual configuration. Matching degree analysis can quantify the degree of fit between content and channels, providing a reliable numerical basis for subsequent dependency assessment and redundancy screening. Compared with fixed binding, this scheme can adapt to the dynamic changes of content and channels, improving the real-time performance and robustness of the association. This application quantitatively characterizes the transmission risk of channel failures to content quality by statistically analyzing the temporal overlap between content anomalies and channel anomalies in historical fault logs, reflecting a causal relationship rather than a simple correlation. Using a time window to determine associated anomalies can tolerate a certain degree of transmission delay and timing error, improving the accuracy of statistics. The calculation results of the fault transmission probability directly reflect the sensitivity of content to channel failures, providing important historical evidence for dependency assessment. By constructing a co-occurrence counting matrix, the historical co-occurrence patterns of the static attributes of content and channels are digitized, utilizing big data statistical patterns. By statistically analyzing and weighting the combination of multiple attribute dimensions, this application captures the varying importance of different attribute pairs. For example, the co-occurrence of bitrate and bandwidth is more critical than the co-occurrence of category and operator, improving the granularity of dependency assessment. This application selects candidate channels based on a comprehensive dependency coefficient ranking, ensuring a good fit between the candidate channels and the content, avoiding blind selection. Constructing a decay line and calculating its slope based on the health score sequence predicts the health trend of channels, providing more forward-looking insights than simply using current health values. Introducing reserved bandwidth average analysis considers resource competition among multiple programs, avoiding the risk of candidate channels being unusable due to bandwidth occupation during actual failures. The availability confidence score integrates remaining available bandwidth and health trends, setting confidence thresholds and exclusive reservation checks. The selected effective redundancy set has higher actual availability, reducing the probability of failover failures. After a failure occurs, this application quickly identifies all content with flexible associations to the failed channel, generating a list of affected content, ensuring complete identification of the failure's impact scope. Analyzing business weights and redundancy coverage gaps quantifies the urgency of recovery for each piece of content. The calculation of the total emergency repair urgency index aggregates the urgency of multiple affected components into a single overall indicator, facilitating maintenance personnel's assessment of the overall severity of the fault. This provides a clear numerical basis for subsequent tiered operations, achieving an upgrade from "qualitative judgment" to "quantitative decision-making." Attached Figure Description
[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1This is a flowchart illustrating the digital television data analysis method based on multi-source data of the present invention. Figure 2 This is a schematic diagram of the structure of the digital television data analysis system based on multi-source data according to the present invention. Detailed Implementation
[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0019] Please see Figure 1 The present invention provides a technical solution: a digital television data analysis method based on multi-source data, the method comprising the following steps: S1: Collect audiovisual content data and transmission channel data, analyze the feature sequences of audiovisual content data and transmission channel data; analyze the correlation based on the matching degree of feature sequences; S1 includes the following: Acquire audiovisual content data and transmission channel data; For audiovisual content data, a frame is extracted every few seconds. Each extracted frame is scaled to a grayscale image of a predetermined size of pixels. The sign of the difference between the grayscale value of each pixel and the average grayscale value of the entire image is calculated. A binary sequence is generated based on the sign of the difference of all pixels as the visual fingerprint at the corresponding moment. Several visual fingerprints are continuously collected to form a content fingerprint sequence. For transmission channel data, a transmission packet is captured every few seconds. The instantaneous bit rate, clock reference jitter value and retransmission request count are extracted from each transmission packet. The three indicators are normalized and combined into a three-dimensional vector as the time delay fingerprint at the corresponding moment. Several time delay fingerprints are continuously collected to form a channel fingerprint sequence. The position of the fixed content fingerprint sequence is used to slide the channel fingerprint sequence forward and backward on the time axis, with the total sliding length being a preset constant. For each offset, the fingerprint similarity of the two sequences at each corresponding position in the overlapping time period after the offset is calculated, and the average value is taken as the matching degree corresponding to the offset. The fingerprint similarity is analyzed by cosine similarity between the two sequences. Take the maximum matching degree under all offsets as the best matching degree, and record the optimal offset corresponding to the best matching degree; Obtain the load rate of the current transmission channel, which is the ratio of the current throughput to the rated capacity; denot the sum of the basic correlation threshold and the modified correlation threshold as the dynamic correlation threshold, which is equal to the product of the basic correlation threshold and the load rate; If the best matching degree is greater than the dynamic association threshold, it is determined that there is a flexible association relationship between the audiovisual content data and the transmission channel data, and the optimal offset and best matching degree corresponding to the flexible association relationship are recorded. For the same audiovisual content data, all flexible relationships are aggregated to generate a relational tuple; the relational tuple includes: audiovisual content data, the content fingerprint sequence corresponding to the audiovisual content data, all transmission channels associated with the audiovisual content data, the corresponding set of optimal offsets and the corresponding set of best matching degrees. S2: Statistically analyze the abnormal time overlap of audiovisual content data and transmission channel data in historical faults; combine the co-occurrence frequency and matching degree of the attributes of audiovisual content data and transmission channel data to analyze the dependency of audiovisual content data and transmission channel data. S2 includes the following: For any associated tuple, extract the occurrence time of all content abnormal events from the audiovisual content data of the associated tuple from the historical fault log. Content abnormal events include, but are not limited to, screen freeze, audio-visual desynchronization, and mosaic burst. Extract the occurrence time of all channel abnormal events from the transmission channel data of the associated tuple from the historical fault log. Channel abnormal events include, but are not limited to, bit rate drop, clock reference jump, and retransmission request count surge. For each content abnormal event, if there is at least one channel abnormal event within a time window of several seconds before and after the occurrence time, then the content abnormal event and the channel abnormal event are counted as one associated abnormality. Count the number of associated anomalies and the total number of channel anomalies in the associated tuples; if the total number of channel anomalies is greater than zero, the ratio of the number of associated anomalies to the total number of channel anomalies is recorded as the fault propagation probability; if the total number of channel anomalies is equal to zero, the fault propagation probability is zero. Extract the static attributes of each audiovisual content data and each transmission channel data separately; The static attributes of audiovisual content data include bitrate level, screen size, and content category; bitrate level includes but is not limited to low, medium, and high; screen size includes but is not limited to standard definition, high definition, and ultra-high definition; content category includes but is not limited to news, sports events, dramas, and variety shows; Static attributes of transmission channel data include, but are not limited to, bandwidth level, carrier, and coverage area; A co-occurrence count matrix is constructed based on the static attributes of audiovisual content data and transmission channel data. The row index of the co-occurrence count matrix corresponds to each specific attribute in the static attribute category of the audiovisual content data, and the column index corresponds to each specific attribute in the static attribute category of the transmission channel data. The matrix elements represent the number of times the static attributes of the audiovisual content data and the static attributes of the transmission channel data co-occur in all historical flexible association relationships. For the current content channel pair, the co-occurrence count under each attribute dimension combination is queried. The co-occurrence counts corresponding to each attribute dimension combination are normalized and then weighted and summed, which is recorded as the comprehensive co-occurrence dependency. The normalized fault propagation probability, the comprehensive co-occurrence dependency, and the best matching degree are weighted and summed to form the comprehensive dependency coefficient. The transmission channel with the largest comprehensive dependency coefficient is used as the primary transmission channel for audiovisual content data.
[0020] S3: Based on the dependency between audiovisual content data and transmission channel data, generate a set of candidate transmission channels for audiovisual content data; analyze the availability confidence of candidate transmission channels when the primary transmission channel fails, and filter to generate an effective redundancy set and a failed redundancy set; S3 includes the following: For the audiovisual content data of the determined primary transmission channel, select the top few transmission channels from the remaining transmission channels other than the primary transmission channel according to the comprehensive dependency coefficient from high to low, and generate a set of candidate transmission channels for the audiovisual content data. Obtain the health score sequence of each candidate transmission channel for the most recent few days, with the number of days as the independent variable and the health score as the dependent variable, and use linear regression to fit the score decay line of each candidate transmission channel to generate the slope of the score decay line. Obtain the average reserved bandwidth of the candidate transmission channel within the historical time window that is occupied by other audiovisual content data or tasks; analyze the remaining available bandwidth of the candidate transmission channel; the remaining available bandwidth is equal to the rated bandwidth of the candidate transmission channel minus the current real-time occupied bandwidth minus the average reserved bandwidth. Analyze the availability confidence of the backup transmission channel when the primary transmission channel fails; Example 1: In this example, the available confidence level = min(1, remaining available bandwidth / (Req+B)) × (1−max(0,−Slope) / A); where Req is the bandwidth required for the audiovisual content data, B represents the zero-prevention constant; Slope represents the slope of the corresponding score decay line; and A represents the preset health decay penalty factor. For each candidate transmission channel, if the availability confidence is greater than or equal to the availability confidence threshold, and the candidate transmission channel is not currently reserved exclusively by audiovisual content data with higher weight, then the candidate transmission channel is included in the effective redundancy set. The fact that the candidate transmission channel is not currently reserved exclusively by audiovisual content data with higher weight means that there is no exclusive reservation from other audiovisual content data with higher priority than the current audiovisual content data in the resource reservation snapshot of the candidate transmission channel. The resource reservation snapshot is used to record the total amount of reserved bandwidth for each candidate transmission channel that has been reserved by other audiovisual content data or tasks in the future time window. Otherwise, the alternative transmission channels will be included in the set of failed redundancies.
[0021] S4: Generate a list of affected content and the service weight of each audiovisual content data based on the faulty transmission channel; integrate fault data to analyze the urgency index of emergency repair, and generate graded alarm signals and switching operation sequences based on the emergency repair urgency threshold.
[0022] S4 includes the following: Obtain the faulty transmission channel and the time of the fault; find all related tuples that have flexible relationships with the faulty channel, extract the audio-visual content data in each related tuple, and generate a list of affected content; analyze the business weight of each audio-visual content data in the list of affected content. Example 2: In this example, the business weight of audiovisual content data is analyzed based on the real-time viewer ratio, live broadcast identifier, and ad insertion identifier. The real-time viewer ratio is equal to the ratio of the number of viewers of the current audiovisual content data to the total number of viewers of all audiovisual content data. If the current audiovisual content data is live, the live broadcast identifier is 1; otherwise, it is 0. If there is an ad insertion in the current audiovisual content data within the current time window, the ad insertion identifier is 1; otherwise, it is 0. The real-time viewer ratio, live broadcast identifier, and ad insertion identifier are weighted and fused to generate the business weight. Analyze the redundancy coverage gap of each audiovisual content data in the affected content list; the redundancy coverage gap is equal to the ratio of the number of candidate transmission channels in the failed redundancy set to the number of candidate transmission channels in the candidate transmission channel set; Analyze the sum of the emergency repair urgency index of audiovisual content data in the list of affected content; the emergency repair urgency index is equal to the product of the business weight of audiovisual content data, the redundancy coverage gap, and the interruption duration amplification factor; the interruption duration amplification factor is equal to the ratio of the duration from the time of the fault to the current time to the preset duration factor. If the calculated interrupt duration amplification factor is greater than the interrupt duration amplification factor threshold, then the corresponding interrupt duration amplification factor is fixed at the interrupt duration amplification factor threshold. Based on the sum of the urgency index of the emergency repairs, tiered operations will be implemented: If the total emergency repair urgency index is less than the first threshold for emergency repair urgency, a blue alert level will be triggered, which will only record the event without triggering any action. If the sum of the emergency repair urgency indices is greater than or equal to the first emergency repair urgency threshold but less than the second emergency repair urgency threshold, a yellow alarm level will be triggered. Select the candidate transmission channel with the highest available confidence from the corresponding effective redundancy set and match it with the corresponding audio-visual content data to generate a one-click switching operation sequence. The audio-visual content data in the one-click switching operation sequence are sorted in descending order from high to low according to the product of business weight and redundancy coverage gap, and pushed to the operation and maintenance console. If the repair urgency index is greater than or equal to the second repair urgency threshold, a red alarm level is triggered, along with an audible and visual alarm. A heatmap is displayed on the large screen showing the product of the redundancy coverage gap and the business weight of each audiovisual content data. A sequence of switching operations is output, arranged in descending order from high to low according to the product of the business weight and the redundancy coverage gap of the audiovisual content data. For audiovisual content data with an empty effective redundancy set, the output switching operation instruction includes a suggestion to reduce the bitrate or switch to backup content.
[0023] Please see Figure 2 The present invention provides a technical solution: a digital television data analysis system based on multi-source data, which includes a correlation analysis module, a transmission dependency analysis module, a candidate screening module, and a rescue index analysis module; The correlation analysis module is used to collect audiovisual content data and transmission channel data, analyze the feature sequences of audiovisual content data and transmission channel data, and analyze correlations based on the matching degree of feature sequences. The transmission dependency analysis module is used to statistically analyze the abnormal time overlap between audiovisual content data and transmission channel data in historical faults; and to analyze the dependency between audiovisual content data and transmission channel data by combining the co-occurrence frequency and matching degree of the attributes of audiovisual content data and transmission channel data. The candidate filtering module is used to generate a set of candidate transmission channels for audiovisual content data based on the dependency between audiovisual content data and transmission channel data; analyze the availability confidence of candidate transmission channels when the primary transmission channel fails, and filter to generate an effective redundancy set and a failed redundancy set. The emergency response index analysis module is used to generate a list of affected content and the business weight of each audiovisual content data based on the faulty transmission channel; it integrates fault data to analyze the emergency response urgency index and generates graded alarm signals and switching operation sequences based on the emergency response urgency threshold.
[0024] The association analysis module includes a fingerprint feature analysis unit and an association matching unit; The fingerprint feature analysis unit is used to acquire audiovisual content data and transmission channel data; and to generate content fingerprint sequences and channel fingerprint sequences based on the audiovisual content data and transmission channel data, respectively. The association matching unit is used to analyze the matching degree between the content fingerprint sequence and the channel fingerprint sequence, and to determine the association relationship between the audiovisual content data and the transmission channel data based on the matching degree. The transport dependency analysis module includes a dependency feature analysis unit and a main transport analysis unit; The dependency feature analysis unit is used to analyze the fault propagation probability and overall co-occurrence dependency of audiovisual content data and data of each transmission channel; The primary transmission analysis unit is used to weight and sum the normalized fault propagation probability, comprehensive co-occurrence dependency, and best matching degree, which is denoted as the comprehensive dependency coefficient. The transmission channel with the largest comprehensive dependency coefficient is used as the primary transmission channel for audiovisual content data. The candidate screening module includes a candidate screening unit, an availability confidence analysis unit, and an availability confidence screening unit; The candidate screening unit is used to select the top few transmission channels from the remaining transmission channels other than the primary transmission channel, sorted by comprehensive dependency coefficient from high to low, to generate a candidate transmission channel set for the audiovisual content data. The available confidence analysis unit is used to construct a score decay line based on the health score sequence of the candidate transmission channel and generate the slope of the score decay line; obtain the average amount of reserved bandwidth of the candidate transmission channel for other audiovisual content data or tasks within the historical time window; analyze the remaining available bandwidth of the candidate transmission channel; and analyze the availability confidence of the candidate transmission channel when the primary transmission channel fails based on the remaining available bandwidth and the slope of the score decay line. The available confidence screening unit is used to select each candidate transmission channel. If the available confidence is greater than or equal to the available confidence threshold, and the candidate transmission channel is not currently reserved exclusively by audiovisual content data with higher weight, then the candidate transmission channel is assigned to the effective redundancy set; otherwise, the candidate transmission channel is assigned to the ineffective redundancy set.
[0025] The emergency response indicator analysis module includes an emergency repair characteristic analysis unit, an emergency repair urgency analysis unit, and an early warning unit; The emergency repair feature analysis unit is used to obtain the faulty transmission channel and the time of the fault occurrence; find all related tuples with flexible correlation to the faulty channel, extract the audio-visual content data in each related tuple, and generate a list of affected content; analyze the business weight of each audio-visual content data in the list of affected content; and analyze the redundancy coverage gap of each audio-visual content data in the list of affected content. The emergency repair urgency analysis unit analyzes the sum of the emergency repair urgency indices of audiovisual content data in the list of affected content; The early warning unit is used to perform tiered operations based on the sum of the emergency repair urgency index.
[0026] 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 process, method, article, or apparatus.
[0027] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.
Claims
1. A digital television data analysis method based on multi-source data, characterized in that, The method includes the following steps: S1: Collect audiovisual content data and transmission channel data, analyze the feature sequences of audiovisual content data and transmission channel data; analyze the correlation based on the matching degree of feature sequences; S2: Statistically analyze the abnormal time overlap of audiovisual content data and transmission channel data in historical faults; combine the co-occurrence frequency and matching degree of the attributes of audiovisual content data and transmission channel data to analyze the dependency of audiovisual content data and transmission channel data. S3: Based on the dependency between audiovisual content data and transmission channel data, generate a set of candidate transmission channels for audiovisual content data; analyze the availability confidence of candidate transmission channels when the primary transmission channel fails, and filter to generate an effective redundancy set and a failed redundancy set; S4: Generate a list of affected content and the business weight of each audiovisual content data based on the faulty transmission channel; By integrating fault data analysis to obtain a repair urgency index, and generating graded alarm signals and switching operation sequences based on the repair urgency threshold.
2. The digital television data analysis method based on multi-source data according to claim 1, characterized in that: S1 includes the following: Acquire audiovisual content data and transmission channel data; generate content fingerprint sequences and channel fingerprint sequences based on the audiovisual content data and transmission channel data, respectively; The matching degree between the content fingerprint sequence and the channel fingerprint sequence is analyzed, and the correlation between the audiovisual content data and the transmission channel data is determined based on the matching degree.
3. The digital television data analysis method based on multi-source data according to claim 2, characterized in that... The generation of the content fingerprint sequence and channel fingerprint sequence includes: for the audiovisual content data, extracting a frame every few seconds, scaling each extracted frame to a grayscale image of a predetermined size of pixels, calculating the sign of the difference between the grayscale value of each pixel and the average grayscale value of the entire image, generating a binary sequence based on the sign of the difference between all pixels as the visual fingerprint at the corresponding moment, and continuously collecting several visual fingerprints to form the content fingerprint sequence. For the data in the transmission channel, a transmission packet is captured every few seconds. The instantaneous bit rate, clock reference jitter value and retransmission request count are extracted from each transmission packet. The three indicators are normalized and combined into a three-dimensional vector as the time delay fingerprint at the corresponding moment. Several time delay fingerprints are continuously collected to form a channel fingerprint sequence.
4. The digital television data analysis method based on multi-source data according to claim 2, characterized in that: S2 includes the following: Analyze the fault propagation probability and overall co-occurrence dependency of audiovisual content data and data from each transmission channel; The normalized fault propagation probability, the comprehensive co-occurrence dependency, and the best matching degree are weighted and summed to form the comprehensive dependency coefficient. The transmission channel with the largest comprehensive dependency coefficient is used as the primary transmission channel for audiovisual content data.
5. The digital television data analysis method based on multi-source data according to claim 4, characterized in that: Fault propagation probability analysis includes: For any associated tuple, extract the occurrence time of all content abnormal events in the audiovisual content data of the associated tuple from the historical fault log; extract the occurrence time of all channel abnormal events in the transmission channel data of the associated tuple from the historical fault log. For each content abnormal event, if there is at least one channel abnormal event within a time window of several seconds before and after the occurrence time, then the content abnormal event and the channel abnormal event are counted as one associated abnormality. Count the number of associated anomalies and the total number of channel anomalies in the associated tuples; if the total number of channel anomalies is greater than zero, the ratio of the number of associated anomalies to the total number of channel anomalies is recorded as the fault propagation probability; if the total number of channel anomalies is equal to zero, the fault propagation probability is zero.
6. The digital television data analysis method based on multi-source data according to claim 4, characterized in that: The comprehensive co-occurrence dependency analysis includes: extracting the static attributes of each audiovisual content data and each transmission channel data respectively; constructing a co-occurrence count matrix based on the static attributes of the audiovisual content data and the transmission channel data, where the matrix elements represent the number of times the static attributes of the audiovisual content data and the static attributes of the transmission channel data co-occur in all historical flexible association relationships; for the current content channel pair, querying the co-occurrence count under each attribute dimension combination, normalizing the co-occurrence counts corresponding to each attribute dimension combination, and then weighting and summing them to record the comprehensive co-occurrence dependency.
7. The digital television data analysis method based on multi-source data according to claim 4, characterized in that: In S3, the following is included: For the audiovisual content data of the determined primary transmission channel, from the other transmission channels besides the primary transmission channel, sort them from high to low according to the comprehensive dependency coefficient, select the first few transmission channels, and generate a set of candidate transmission channels for the audiovisual content data. Construct a score decay line based on the health score sequence of the alternative transmission channel to generate the slope of the score decay line; obtain the average reserved bandwidth of the alternative transmission channel for other audiovisual content data or tasks within the historical time window; analyze the remaining available bandwidth of the alternative transmission channel; and analyze the availability confidence of the alternative transmission channel when the primary transmission channel fails based on the remaining available bandwidth and the slope of the score decay line. For each candidate transmission channel, if the available confidence level is greater than or equal to the available confidence level threshold, and the candidate transmission channel is not currently reserved exclusively by audiovisual content data with higher weight, then the candidate transmission channel is included in the effective redundancy set. Otherwise, the candidate transmission channels will be included in the set of failed redundancies.
8. The digital television data analysis method based on multi-source data according to claim 7, characterized in that: S4 includes the following: Obtain the faulty transmission channel and the time of the fault; find all related tuples that have a flexible relationship with the faulty channel, extract the audio-visual content data from each related tuple, and generate a list of affected content; Analyze the business weight of each audiovisual content data in the affected content list; analyze the redundancy coverage gap of each audiovisual content data in the affected content list; Analyze the sum of the emergency repair urgency indexes of audiovisual content data in the affected content list; The tiered operation is performed based on the sum of the emergency repair urgency indexes.
9. The digital television data analysis method based on multi-source data according to claim 8, characterized in that: The specific steps for performing graded operations based on the sum of the emergency repair urgency indices include: if the sum of the emergency repair urgency indices is less than the first threshold of emergency repair urgency, then a blue alert level is triggered, and only the event is recorded without triggering any action; If the sum of the emergency repair urgency indices is greater than or equal to the first emergency repair urgency threshold but less than the second emergency repair urgency threshold, a yellow alarm level will be triggered. Select the candidate transmission channel with the highest available confidence from the corresponding effective redundancy set and match it with the corresponding audio-visual content data to generate a one-click switching operation sequence. The audio-visual content data in the one-click switching operation sequence are arranged in descending order from high to low according to the product of business weight and redundancy coverage gap, and pushed to the operation and maintenance console. If the repair urgency index is greater than or equal to the second repair urgency threshold, a red alarm level is triggered, along with an audible and visual alarm. A heatmap is displayed on the large screen showing the product of the redundancy coverage gap and the business weight of each audiovisual content data. A sequence of switching operations is output, arranged in descending order from high to low according to the product of the business weight and the redundancy coverage gap of the audiovisual content data. For audiovisual content data with an empty effective redundancy set, the output switching operation instruction includes a suggestion to reduce the bitrate or switch to backup content.
10. A digital television data analysis system based on multi-source data, wherein the system is applied to the digital television data analysis method based on multi-source data as described in any one of claims 1-9, characterized in that, The system includes a correlation analysis module, a transport dependency analysis module, a candidate screening module, and a rescue indicator analysis module; The correlation analysis module is used to collect audiovisual content data and transmission channel data, and analyze the feature sequences of audiovisual content data and transmission channel data. Analysis of association relationships based on the matching degree of feature sequences; The transmission dependency analysis module is used to statistically analyze the abnormal time overlap between audiovisual content data and transmission channel data in historical faults; and to analyze the dependency between audiovisual content data and transmission channel data by combining the co-occurrence frequency and matching degree of the attributes of audiovisual content data and transmission channel data. The candidate screening module is used to generate a set of candidate transmission channels for audiovisual content data based on the dependency between audiovisual content data and transmission channel data; analyze the availability confidence of candidate transmission channels when the primary transmission channel fails, and filter to generate an effective redundancy set and a failed redundancy set. The rescue index analysis module is used to generate a list of affected content and the business weight of each audiovisual content data based on the faulty transmission channel. By integrating fault data analysis to obtain a repair urgency index, and generating graded alarm signals and switching operation sequences based on the repair urgency threshold.