E-commerce platform live broadcast signal anomaly detection method and system

Abstract

The invention discloses an e-commerce platform live broadcast signal anomaly detection method and system, and relates to the related technical field of live broadcast signal detection, and the method comprises the steps: connecting a target streaming media platform, and obtaining a plurality of streaming media signal data; performing quality monitoring on the plurality of streaming media signal data, and extracting a plurality of signal quality feature data; inputting the plurality of signal quality characteristic data into a signal anomaly detection model for judgment to obtain signal detection judgment data; and according to the signal detection judgment data, generating a live broadcast signal detection report for visual display. The technical problems that in the prior art, live broadcast signal anomaly detection is poor in real-time performance and insufficient in accuracy, and a complex live broadcast scene cannot be effectively processed, so that the user experience is poor are solved, and the technical effects of improving the real-time performance and the detection precision of live broadcast signal anomaly detection and the user experience are achieved.

Description

Technical Field

[0001] The present application relates to the technical field related to live broadcast signal detection, and specifically to a method and system for detecting abnormalities in live broadcast signals of an e-commerce platform. Background Art

[0002] In recent years, the rapid development of the e-commerce industry has promoted the widespread application of live broadcast technology. Especially in marketing activities, it has become a trend for e-commerce platforms to display products and interactively sell through live broadcasts. However, during the live broadcast process, due to network fluctuations, excessive server load or unstable streaming media signals, abnormal audio and video signals often occur, such as picture freezes, signal interruptions, audio and video asynchrony, etc., which seriously affect the user experience. Traditional live broadcast signal anomaly monitoring is mainly based on simple threshold judgment, which has limitations such as insufficient real-time performance, low accuracy and inability to cope with complex scenarios. It is difficult to accurately and quickly identify and locate anomalies that may occur during the live broadcast process, and no visual detection report is generated, which affects the platform's live broadcast quality optimization and efficient handling of anomalies.

[0003] Therefore, in the current relevant technologies, there are technical problems such as poor real-time and accuracy of live signal anomaly detection and inability to effectively handle complex live broadcast scenarios, resulting in poor user experience. Summary of the invention

[0004] The present application provides a method and system for detecting anomalies in live broadcast signals of an e-commerce platform, thereby solving the technical problems in the prior art of poor real-time and accuracy of live broadcast signal anomaly detection and inability to effectively handle complex live broadcast scenarios, resulting in poor user experience, and achieving the technical effect of improving the real-time performance and detection accuracy of live broadcast signal anomaly detection and user experience.

[0005] The present application provides a method for detecting anomalies in live broadcast signals of an e-commerce platform, the method comprising: connecting to a target streaming media platform to obtain a plurality of streaming media signal data; performing quality monitoring on the plurality of streaming media signal data to extract a plurality of signal quality feature data; inputting the plurality of signal quality feature data into a signal anomaly detection model for judgment to obtain signal detection judgment data; and generating a live broadcast signal detection report based on the signal detection judgment data for visual display.

[0006] In a possible implementation, the method for detecting anomalies in live broadcast signals of an e-commerce platform further performs the following processing: if the signal detection determines that the data is abnormal, an automatic alarm mechanism is triggered and an intelligent repair strategy is generated; wherein the intelligent repair strategy includes switching to a backup server, adjusting bandwidth, and reconnecting to the target streaming media platform; and repair processing is performed according to the intelligent repair strategy.

[0007] In a possible implementation, the method for detecting anomalies in live broadcast signals of an e-commerce platform further performs the following processing: extracting video stream data, audio stream data and network traffic data based on the multiple streaming media signal data; preprocessing the video stream data and the audio stream data to obtain video stream preprocessing data and audio stream preprocessing data, wherein the preprocessing step includes: a: performing timing synchronization according to PTS / DTS of audio and video; b: performing format detection and verification on the video stream data and audio stream data.

[0008] In a possible implementation, the method for detecting anomalies in live broadcast signals of an e-commerce platform further performs the following processing: performing evaluation based on the video stream preprocessing data, the audio stream preprocessing data and the network traffic data, including dynamic bandwidth monitoring, delay measurement, packet loss rate calculation, video frame rate detection, video freeze analysis and audio and video synchronization analysis, to obtain multiple signal evaluation data; performing quality feature extraction on the multiple signal evaluation data to obtain multiple signal quality feature data, including bandwidth, delay, packet loss rate, frame rate, video freeze, and audio and video synchronization.

[0009] In a possible implementation, the method for detecting anomalies in live broadcast signals of an e-commerce platform further performs the following processing: collecting streaming media signal recording data of the target live broadcast platform; labeling and classifying the streaming media signal recording data according to preset quality rules to obtain multiple normal signal recording data and multiple abnormal signal recording data; wherein the preset quality rules include multiple key signal monitoring features; and using the multiple key signal monitoring features, the multiple normal signal recording data and the multiple abnormal signal recording data for training to obtain a signal anomaly detection model.

[0010] In a possible implementation, the method for detecting abnormalities in live broadcast signals of an e-commerce platform also performs the following processing: if the bandwidth, delay, or packet loss rate exceeds the corresponding bandwidth threshold value, delay threshold value, or packet loss rate threshold value, or the frame rate is lower than the normal frame rate range, or the audio and video synchronization is lower than a preset threshold, and the video screen is frozen, it is judged as abnormal signal recording data.

[0011] In a possible implementation, the method for detecting abnormality of live broadcast signal on an e-commerce platform further performs the following processing: screening abnormal signal data in the signal detection judgment data to obtain the abnormal probability value output by the signal abnormality detection model; determining the quality characteristics corresponding to the abnormal signal data; comparing the current abnormal signal with the historical abnormal signal to calculate the abnormal signal matching degree; performing confidence scoring on the abnormal signal data, and classifying the abnormal signal according to the confidence scoring result; the confidence scoring formula is as follows:

[0012]

[0013] Among them, A is the confidence score, P model represents the abnormal probability value, F i represents the normalized value of the i-th feature of the abnormal signal, w i represents the weight of the i-th feature, S history represents the abnormal signal matching degree, α, β, and γ represent weight parameters.

[0014] The present application also provides an e-commerce platform live signal anomaly detection system, including: a streaming media signal data acquisition module, used to connect to a target streaming media platform and acquire multiple streaming media signal data; a signal quality feature data extraction module, used to monitor the quality of the multiple streaming media signal data and extract multiple signal quality feature data; a signal detection judgment data acquisition module, used to input the multiple signal quality feature data into a signal anomaly detection model for judgment and obtain signal detection judgment data; a live signal detection report generation module, used to generate a live signal detection report for visual display based on the signal detection judgment data.

[0015] The present application proposes a method and system for detecting anomalies in live broadcast signals on an e-commerce platform, connects to a target streaming media platform, obtains multiple streaming media signal data; performs quality monitoring on multiple streaming media signal data, extracts multiple signal quality feature data; inputs multiple signal quality feature data into a signal anomaly detection model for judgment, obtains signal detection judgment data; generates a live broadcast signal detection report for visual display based on the signal detection judgment data. This solves the technical problems in the prior art of poor real-time and accuracy of live broadcast signal anomaly detection, and the inability to effectively handle complex live broadcast scenes, resulting in a poor user experience, and achieves the technical effect of improving the real-time and detection accuracy of live broadcast signal anomaly detection and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solution of the embodiment of the present disclosure, the accompanying drawings of the embodiment of the present disclosure will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0017] Figure 1 A schematic diagram of a method for detecting abnormality of a live broadcast signal on an e-commerce platform provided in an embodiment of the present application;

[0018] Figure 2A schematic diagram of the structure of an anomaly detection system for live broadcast signals of an e-commerce platform provided in an embodiment of the present application.

[0019] Explanation of the reference numerals: streaming media signal data acquisition module 10, signal quality feature data extraction module 20, signal detection and judgment data acquisition module 30, live signal detection report generation module 40. DETAILED DESCRIPTION

[0020] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0021] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0022] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0023] The present application embodiment provides a method for detecting abnormality of live broadcast signal of an e-commerce platform, such as Figure 1 As shown, the method includes:

[0024] Step S100, connecting to a target streaming media platform and acquiring a plurality of streaming media signal data.

[0025] Preferably, connecting to the target streaming media platform refers to connecting to the live broadcast server of the platform through a standardized interface (API) or streaming media protocol to obtain signal data transmitted during the live broadcast process, wherein the target streaming media platform refers to a specific e-commerce platform or live broadcast service platform, such as Taobao Live, Douyin Live, Kuaishou Live, etc., or a platform using a public streaming media protocol (such as RTMP, HLS), and then obtaining multiple streaming media signal data, including video stream data, audio stream data and network traffic data. The video stream is real-time data transmitted through a streaming media protocol (such as RTMP or HTTP-FLV) and is used to constitute a live broadcast picture. The video stream data includes video encoding information of the live broadcast picture, such as Frame rate, resolution, bit rate, key frame interval (GOP), etc.; audio stream is synchronously transmitted audio signal data, which is used to constitute the sound part of the live broadcast. The audio stream data includes the audio encoding information in the live broadcast, such as sampling rate, audio bit rate, number of audio channels, etc.; network traffic data refers to the network transmission-related data generated during the live broadcast, including but not limited to bandwidth usage (the amount of data transmitted per second), delay (the time difference from sending to receiving data), packet loss rate (the percentage of data packets lost during network transmission to the total transmission packets), etc. By obtaining and extracting this data from the platform, it is possible to analyze the live broadcast quality and detect possible abnormal situations, such as picture freezes, audio distortion or signal interruptions.

[0026] Step S200: monitoring the quality of the plurality of streaming media signal data and extracting a plurality of signal quality feature data.

[0027] Preferably, by analyzing the acquired streaming media signal data in real time, important parameters reflecting the signal quality are extracted so as to conduct a comprehensive quality assessment of the live signal. Specifically, the real-time transmitted streaming media data (video stream, audio stream and network traffic data) is analyzed to ensure the stability and reliability of signal transmission, identify potential problems that may affect the live broadcast effect (such as freeze, delay, packet loss, audio and video asynchrony, etc.), and then extract signal quality feature data. The video stream quality feature data may include but is not limited to frame rate (the number of frames transmitted per second in the video, reflecting the smoothness of the picture, low frame rate may cause picture freeze), resolution (pixel size of the video picture, such as 1920x1080, low resolution may cause blurred picture), video bit rate (the amount of data transmitted per second in the video) and key frame interval (GOP, the interval time between two key frames, affecting the video The decoding and picture quality recovery capabilities of the video stream, too large an interval may cause a delay in picture recovery); audio stream quality feature data may include but are not limited to sampling rate (the number of audio signals sampled per second, such as 44.1kHz or 48kHz, a low sampling rate may cause audio distortion), audio bit rate (the amount of data transmitted per second for audio, a low bit rate may cause a decline in sound quality), audio and video synchronization (whether the audio and video signals are transmitted synchronously, if the audio is ahead of or behind the video, it will affect the user experience); network traffic quality feature data may include but are not limited to bandwidth (insufficient bandwidth may cause data transmission interruption or quality degradation), latency (the time required for data to travel from the sender to the receiver, high latency may cause picture lag or interaction delay) and packet loss rate (the ratio of the number of data packets lost during transmission to the total number of data packets, a high packet loss rate will cause the picture or audio to be intermittent). Through the extraction and analysis of quality feature data, the platform can discover and solve signal anomalies in live broadcasts in real time to ensure that users enjoy a smooth live broadcast experience.

[0028] Furthermore, step S200 also includes step S210, extracting video stream data, audio stream data and network traffic data based on the multiple streaming media signal data; step S220, preprocessing the video stream data and the audio stream data to obtain video stream preprocessing data and audio stream preprocessing data, wherein the preprocessing step includes: a: performing timing synchronization according to PTS / DTS of audio and video; b: performing format detection and verification on the video stream data and audio stream data.

[0029] Preferably, video stream data, audio stream data and network traffic data are extracted respectively from multiple streaming media signal data. Video stream data refers to the encoding information of the video picture, such as frame rate (FPS), resolution, bit rate, etc., which is used to analyze the smoothness and clarity of the picture; audio stream data refers to the encoding information of the sound part, such as sampling rate, audio bit rate, channel information, etc. Used to analyze sound quality and synchronization; network traffic data includes network transmission related parameters, such as bandwidth, delay, packet loss rate, etc., which are used to evaluate the stability and efficiency of network transmission; pre-processing of video stream data and audio stream data, including timing synchronization according to PTS / DTS of audio and video, where PTS represents the presentation timestamp of video frame or audio frame, which is used to indicate when the frame is displayed on the screen or played in the speaker, and DTS represents the decoding timestamp of video frame or audio frame, which is used to indicate when the frame is decoded in the decoder. In live broadcast, audio and video are usually encoded and transmitted independently, which may lead to asynchrony due to network delay or different decoding speeds, such as the picture cannot keep up with the sound or the sound lags behind the picture. Use PTS / DTS information to adjust the playback order of audio frames and video frames to ensure that they are displayed synchronously at the same time point, or adjust the decoder processing order according to the decoding time (DTS), and adjust the playback order according to the presentation time (PTS).

[0030] Preferably, format detection and verification are performed on the video stream data and the audio stream data. During the transmission process, the video and audio streams may have format errors due to inconsistent encoder settings or data corruption. For example, the video frame encoding format (such as H.264, H.265) is incompatible with the decoder, or the audio sampling rate does not meet the requirements. Specifically, the video stream format detection includes encoding format verification, checking whether the video stream uses a supported encoding format (such as H.264, VP8, etc.), resolution detection, verifying whether the resolution meets the expected range (such as 720p, 1080p), and frame rate detection, confirming whether the frame rate meets the expected range (such as 4K). 30fps or 60fps); audio stream format detection includes sampling rate verification, checking whether the audio sampling rate meets the standard (such as 44.1kHz or 48kHz), audio encoding format verification, checking whether the audio uses a supported encoding format (such as AAC, MP3), channel detection, confirming whether the audio is mono, stereo or multi-channel, and meeting the live broadcast requirements, etc., and then obtaining video stream preprocessing data and audio stream preprocessing data to ensure the compatibility of streaming media data, avoid decoding failure or live broadcast interruption due to format errors, provide high-quality audio and video transmission, and meet the needs of different devices and scenarios.

[0031] Furthermore, step S200 also includes step S230, performing evaluation based on the video stream preprocessing data, the audio stream preprocessing data and the network traffic data, including dynamic bandwidth monitoring, delay measurement, packet loss rate calculation, video frame rate detection, video freeze analysis and audio and video synchronization analysis, to obtain multiple signal evaluation data; step S240, performing quality feature extraction on the multiple signal evaluation data to obtain multiple signal quality feature data, including bandwidth, delay, packet loss rate, frame rate, video freeze, and audio and video synchronization.

[0032] Preferably, the evaluation is performed based on the video stream preprocessing data, audio stream preprocessing data and network traffic data, specifically, including dynamic bandwidth monitoring, delay measurement, packet loss rate calculation, video frame rate detection, video freeze analysis and audio and video synchronization analysis. Dynamic bandwidth monitoring refers to real-time analysis of the data packet transmission rate of the network layer, monitoring the actual bandwidth used during the live broadcast, dynamically recording the changes in bandwidth over time, and judging whether the minimum transmission requirements of the video and audio streams are met, such as blurred live broadcast images or signal interruptions due to network congestion; delay measurement refers to evaluating the total transmission delay by calculating the time difference between the audio and video data packets sent from the source to the receiving end. A higher delay may cause the live broadcast image to lag, affecting the audience's real-time interactive experience; packet loss rate calculation is to calculate the ratio of the number of lost data packets to the number of sent data packets. A higher packet loss rate will cause video freezes, audio distortion or signal interruptions; video frame rate detection is to record the decoded data through video stream preprocessing data. The frame count is used to count the number of frames played per second, compare the actual frame rate with the expected frame rate (such as 30FPS or 60FPS), and detect whether the frame rate is stable. For example, insufficient or fluctuating frame rate may cause the picture to freeze or jitter. Video freeze analysis is to detect whether the interval between frames exceeds the normal range and analyze the fluctuation of video bit rate. When the bit rate drops sharply, freeze may occur and determine the cause of freeze (such as network problems or insufficient encoder performance). Audio and video synchronization analysis is to compare the time difference between audio and video based on the PTS of the two, detect whether the decoding and output time of audio and video are consistent during playback, improve the synchronization of audio and video, eliminate the phenomenon of audio and video being out of sync, and enhance the user's viewing experience. Then, key signal quality feature data is extracted from the evaluation results for anomaly detection and quality optimization of live signal, including bandwidth, delay, packet loss rate, frame rate, video freeze, and audio and video synchronization, which helps to ensure the stability and high-quality output of live signal.

[0033] Step S300: input the plurality of signal quality feature data into a signal anomaly detection model for judgment to obtain signal detection judgment data.

[0034] Preferably, the extracted signal quality feature data is input into the signal anomaly detection model to realize automated and precise signal quality analysis and anomaly identification, wherein the signal anomaly detection model is an anomaly identification classification model based on random forest, which is used to analyze the input signal quality feature data and determine whether the signal is in a normal state or abnormal. Specifically, multiple signal quality feature data (such as bandwidth, delay, packet loss rate, frame rate, video jamming data and audio synchronization) are input into the signal anomaly detection model for analysis and processing, and the signal detection judgment data is output, which is divided into normal signal data (the detection feature data are all within the normal range, and the signal quality is normal) and abnormal signal data (one or more of the detection feature data exceeds the threshold range, and the signal is judged to be abnormal). The specific type of signal anomaly is also output, for example, video jamming, abnormal frame rate fluctuation or insufficient bit rate, audio distortion, audio bit rate is too low or audio and video are not synchronized, network anomaly, high delay, high packet loss rate or insufficient bandwidth. A fully automated analysis process from signal quality feature data to anomaly detection is realized, and the efficiency and accuracy of anomaly identification are significantly improved through an intelligent anomaly detection model, which helps to ensure the stability of live broadcast and high-quality user experience.

[0035] Furthermore, step S300 also includes step S310, if the signal detection determines that the data is abnormal, triggering an automatic alarm mechanism and generating an intelligent repair strategy; step S320, wherein the intelligent repair strategy includes switching the backup server, adjusting the bandwidth, and reconnecting to the target streaming media platform; step S330, performing repair processing according to the intelligent repair strategy.

[0036] Preferably, when an abnormality is detected in the live broadcast signal, a response mechanism is automatically activated to solve the problem and restore the normal state of the live broadcast signal through alarm and repair strategies. Specifically, when an abnormal signal is detected (such as high latency, high packet loss rate, low frame rate, etc.), an alarm is automatically triggered to notify the administrator of the detailed information of the abnormal situation to ensure that the problem is discovered in time and measures are taken to reduce the interruption time of the live broadcast, and an intelligent repair strategy (solution) is automatically generated based on the type and condition of the signal abnormality, and the live broadcast signal is repaired, including switching backup servers, adjusting bandwidth, and reconnecting to the target streaming platform. That is, when it is detected that the target server is overloaded or the network connection is unstable, the status of all backup servers is checked. The system automatically switches to the pre-configured backup server based on the current state (such as load conditions, network connection stability, etc.), gives priority to the backup server with the best status, and redirects the current live stream to the server; when insufficient network bandwidth is detected, the transmission bit rate of the streaming media is dynamically adjusted to adapt to the current network conditions, such as reducing the video bit rate (for example, from 1080p to 720p or 480p) to reduce the amount of transmitted data; when the live connection is interrupted or unstable, it automatically tries to reconnect to the streaming platform to restore the normal transmission of the live stream, such as detecting the current connection status and recording the reason for the disconnection (such as network timeout, authentication failure, etc.), automatically initiating a reconnection request, and re-establishing the transmission link with the target streaming platform.

[0037] Furthermore, step S300 also includes step S340, collecting streaming signal recording data of the target live broadcast platform; step S350, labeling and classifying the streaming signal recording data according to preset quality rules to obtain multiple normal signal recording data and multiple abnormal signal recording data; step S360, wherein the preset quality rules include multiple key signal monitoring features; step S370, using the multiple key signal monitoring features, the multiple normal signal recording data and the multiple abnormal signal recording data for training to obtain a signal anomaly detection model.

[0038] Step S350 further includes: if the bandwidth, delay, packet loss rate exceeds the corresponding bandwidth threshold, delay threshold, packet loss rate threshold, or the frame rate is lower than the normal frame rate range, or the audio and video synchronization is lower than the preset threshold and the video screen is frozen, it is judged as abnormal signal recording data.

[0039] Preferably, historical streaming media signal record data is collected from the log data of the target live broadcast platform, including records of audio, video and network traffic information transmitted during the live broadcast, and then the streaming media signal record data is labeled and classified according to preset quality rules, wherein the preset quality rules refer to quality rules for judging the state of streaming media signals, including multiple key signal monitoring features and their threshold values ​​and conditions, describing the definition criteria of normal signals and abnormal signals, if the bandwidth exceeds the bandwidth threshold value (such as less than 500kbps), the delay exceeds the delay threshold value (such as higher than 200ms), the packet loss rate exceeds the packet loss rate threshold value (such as more than 5%), the frame rate is lower than the normal range of the frame rate (such as less than 24FPS or fluctuates frequently), the audio and video synchronization (such as the difference between the audio and video playback time exceeds 50ms) and the video picture freezes (such as the freeze lasts for more than 3 seconds or the frequency of occurrence is too high), it is judged as abnormal signal record data, and then the abnormal signal record data is processed. The streaming media signal recording data is labeled and classified into normal signal recording data (meeting all threshold conditions and stable signal quality) and abnormal signal recording data (at least one feature exceeds the threshold range and the signal has quality problems); then, multiple key signal monitoring features (such as bandwidth, delay, packet loss rate, etc.) are used as data features, multiple normal signal recording data are used as positive samples, and multiple abnormal signal recording data are used as negative samples. A classification model is constructed based on random forest, and the key signal monitoring features are used as input variables, and the signal status (normal or abnormal) is used as the output label. The classification model is trained to obtain a signal anomaly detection model, which can automatically analyze the input streaming media signal feature data, determine whether the signal is abnormal in real time, and output detection results (such as "normal signal" or "abnormal signal"), so as to quickly identify and respond to signal abnormality problems, improve user viewing experience, and improve the accuracy and automation of live signal monitoring.

[0040] Further, step S300 also includes step S380, screening the abnormal signal data in the signal detection judgment data to obtain the abnormal probability value output by the signal abnormality detection model; step S390, determining the quality characteristics corresponding to the abnormal signal data; step S3100, comparing the current abnormal signal with the historical abnormal signal, and calculating the abnormal signal matching degree; step S3200, performing confidence scoring on the abnormal signal data, and classifying the abnormal signal according to the confidence scoring result; step S3300, the confidence scoring formula is as follows:

[0041]

[0042] Among them, A is the confidence score, P model represents the abnormal probability value, F i represents the normalized value of the i-th feature of the abnormal signal, w i represents the weight of the i-th feature, S historyrepresents the abnormal signal matching degree, α, β, and γ represent weight parameters.

[0043] Preferably, the abnormal signal data in the signal detection judgment data is screened, and the abnormal probability value corresponding to each signal is output. The abnormal probability value indicates the possibility of a certain signal being judged as abnormal, and the range is usually 0 to 1. Only the signal data with a high abnormal probability value (such as exceeding the set threshold) is screened as the abnormal signal data; then the quality characteristics corresponding to the abnormal signal data (such as signal quality parameters such as bandwidth, delay, packet loss rate, frame rate, and analysis results such as video freeze and audio and video synchronization) are extracted, and the characteristics of the current abnormal signal are compared with those of the historical abnormal signals, and the matching degree is calculated to indicate the similarity between the current abnormality and the historical abnormality, and the range is 0 to 1, which is used to determine whether the current abnormality is associated with a known problem; then according to the confidence scoring formula Calculate the confidence score of the abnormal signal, which represents the comprehensive credibility of the abnormal signal and is used to measure the severity and credibility of the abnormality. model represents the abnormal probability value, F i represents the normalized value of the i-th feature of the abnormal signal, w i represents the weight of the i-th feature, S history represents the matching degree of the abnormal signal, α, β, and γ represent weight parameters; finally, different levels are set according to the range of the confidence score A. For example, an abnormal signal with a confidence score A>0.8 is a high-priority abnormality (needs to be processed immediately), an abnormal signal with a confidence score A≤0.8 is a medium-priority abnormality (recommended to be processed as soon as possible), and an abnormal signal with a confidence score A<0.5 is a low-priority abnormality (can be processed later), which helps to improve the efficiency of abnormality detection and management.

[0044] Step S400, generating a live signal detection report for visual display based on the signal detection judgment data.

[0045] Preferably, generating a live signal detection report based on signal detection judgment data refers to generating multi-dimensional signal judgment data for evaluating the quality of the live signal, which may include signal status (whether there is signal loss, signal strength change curve at different time points), audio quality (video clarity, audio distortion, noise detection), network performance (packet loss rate, jitter and delay change trend), abnormal conditions (signal interruption time and duration, video screen distortion, audio abnormality and other events, the number of occurrences and time periods) and causes of signal problems (such as freezes caused by network packet loss, and delays caused by excessive encoding bit rate); visual display is a graphical expression of the detection report to help users quickly understand the status and problems of the live signal, including displaying the overall results of the signal quality score, the proportion of different signal states (such as normal, freezes, and loss), real-time signal strength change curves, and statistics on the number of occurrences of various abnormal events, etc. Users can click to view detailed signal quality analysis within a certain period of time, and provide problem location function. For example, clicking on an abnormal event can view the corresponding video frame or network status.

[0046] In the above, refer to Figure 1 A method for detecting abnormality of live broadcast signals of an e-commerce platform according to an embodiment of the present invention is described in detail. Figure 2 A system for detecting anomalies in live broadcast signals of an e-commerce platform according to an embodiment of the present invention is described.

[0047] According to an embodiment of the present invention, a live broadcast signal anomaly detection system for an e-commerce platform is used to solve the technical problems of poor real-time and accuracy of live broadcast signal anomaly detection in the prior art, and the inability to effectively handle complex live broadcast scenes, resulting in poor user experience, and achieves the technical effect of improving the real-time and detection accuracy of live broadcast signal anomaly detection and user experience. An e-commerce platform live broadcast signal anomaly detection system includes: a streaming media signal data acquisition module 10, a signal quality feature data extraction module 20, a signal detection judgment data acquisition module 30, and a live broadcast signal detection report generation module 40.

[0048] The streaming media signal data acquisition module 10 is used to connect to the target streaming media platform and obtain multiple streaming media signal data; the signal quality feature data extraction module 20 is used to monitor the quality of the multiple streaming media signal data and extract multiple signal quality feature data; the signal detection judgment data acquisition module 30 is used to input the multiple signal quality feature data into the signal anomaly detection model for judgment and obtain signal detection judgment data; the live signal detection report generation module 40 is used to generate a live signal detection report for visual display based on the signal detection judgment data.

[0049] The specific configuration of the signal detection judgment data acquisition module 30 will be described in detail below. The signal detection judgment data acquisition module 30 further includes: if the signal detection judgment data is abnormal, triggering an automatic alarm mechanism and generating an intelligent repair strategy; wherein the intelligent repair strategy includes switching a backup server, adjusting bandwidth, and reconnecting to the target streaming media platform; and performing repair processing according to the intelligent repair strategy.

[0050] The specific configuration of the signal quality feature data extraction module 20 will be described in detail below. The signal quality feature data extraction module 20 further includes: extracting video stream data, audio stream data and network traffic data based on the multiple streaming media signal data; preprocessing the video stream data and the audio stream data to obtain video stream preprocessing data and audio stream preprocessing data, wherein the preprocessing step includes: a: performing timing synchronization according to the PTS / DTS of the audio and video; b: performing format detection and verification on the video stream data and the audio stream data.

[0051] The specific configuration of the signal quality feature data extraction module 20 will be described in detail below. The signal quality feature data extraction module 20 further includes: performing evaluation based on the video stream preprocessing data, the audio stream preprocessing data and the network traffic data, including bandwidth dynamic monitoring, delay measurement, packet loss rate calculation, video frame rate detection, video freeze analysis and audio and video synchronization analysis, to obtain multiple signal evaluation data; performing quality feature extraction on the multiple signal evaluation data to obtain multiple signal quality feature data, including bandwidth, delay, packet loss rate, frame rate, video freeze, and audio and video synchronization.

[0052] The specific configuration of the signal detection judgment data acquisition module 30 will be described in detail below. The signal detection judgment data acquisition module 30 further includes: collecting streaming media signal recording data of the target live broadcast platform; labeling and classifying the streaming media signal recording data according to preset quality rules to obtain multiple normal signal recording data and multiple abnormal signal recording data; wherein the preset quality rules include multiple key signal monitoring features; using the multiple key signal monitoring features, the multiple normal signal recording data and the multiple abnormal signal recording data for training to obtain a signal anomaly detection model.

[0053] The specific configuration of the signal detection judgment data acquisition module 30 will be described in detail below. The signal detection judgment data acquisition module 30 further includes: if the bandwidth, delay, packet loss rate exceeds the corresponding bandwidth threshold value, delay threshold value, packet loss rate threshold value, or the frame rate is lower than the normal frame rate range, or the audio and video synchronization is lower than the preset threshold value and the video screen is stuck, it is judged as abnormal signal recording data.

[0054] The specific configuration of the signal detection judgment data acquisition module 30 will be described in detail below. The signal detection judgment data acquisition module 30 further includes: screening the abnormal signal data in the signal detection judgment data to obtain the abnormal probability value output by the signal abnormality detection model; determining the quality characteristics corresponding to the abnormal signal data; comparing the current abnormal signal with the historical abnormal signal to calculate the abnormal signal matching degree; performing confidence scoring on the abnormal signal data, and classifying the abnormal signal according to the confidence scoring result; the confidence scoring formula is as follows:

[0055]

[0056] Among them, A is the confidence score, P model represents the abnormal probability value, F i represents the normalized value of the i-th feature of the abnormal signal, w i represents the weight of the i-th feature, S history represents the abnormal signal matching degree, α, β, and γ represent weight parameters.

[0057] An e-commerce platform live broadcast signal anomaly detection system provided by an embodiment of the present invention can execute an e-commerce platform live broadcast signal anomaly detection method provided by any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0058] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0059] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for detecting abnormality of live broadcast signal on an e-commerce platform, characterized in that: The method comprises: Connect to the target streaming media platform and obtain multiple streaming media signal data; Performing quality monitoring on the plurality of streaming media signal data and extracting a plurality of signal quality feature data; Inputting the plurality of signal quality feature data into a signal anomaly detection model for judgment to obtain signal detection judgment data; Based on the signal detection judgment data, a live signal detection report is generated for visual display.

2. The method for detecting abnormality of live broadcast signal of an e-commerce platform according to claim 1, characterized in that: The method further comprises: If the signal detection determines that the data is abnormal, an automatic alarm mechanism is triggered and an intelligent repair strategy is generated; The intelligent repair strategy includes switching backup servers, adjusting bandwidth, and reconnecting to the target streaming media platform; Repair processing is performed according to the intelligent repair strategy.

3. The method for detecting abnormality of live broadcast signal of an e-commerce platform according to claim 1, characterized in that: The method further comprises: Extracting video stream data, audio stream data and network traffic data based on the plurality of streaming media signal data; The video stream data and the audio stream data are preprocessed to obtain video stream preprocessed data and audio stream preprocessed data, wherein the preprocessing step includes: a: Timing synchronization based on PTS / DTS of audio and video; b: Perform format detection and verification on the video stream data and audio stream data.

4. The method for detecting abnormality of live broadcast signal of an e-commerce platform according to claim 3, characterized in that: The quality of the plurality of streaming media signal data is monitored to extract a plurality of signal quality feature data, including: Performing evaluation based on the video stream preprocessing data, the audio stream preprocessing data and the network traffic data, including bandwidth dynamic monitoring, delay measurement, packet loss rate calculation, video frame rate detection, video freeze analysis and audio and video synchronization analysis, to obtain multiple signal evaluation data; Quality features are extracted from the multiple signal evaluation data to obtain multiple signal quality feature data, including bandwidth, delay, packet loss rate, frame rate, video freeze, and audio and video synchronization.

5. The method for detecting abnormality of live broadcast signal of an e-commerce platform according to claim 1, characterized in that: Build a signal anomaly detection model, including: Collect streaming media signal recording data of the target live broadcast platform; According to preset quality rules, the streaming media signal recording data is labeled and classified to obtain a plurality of normal signal recording data and a plurality of abnormal signal recording data; Wherein, the preset quality rules include a plurality of key signal monitoring features; The plurality of key signal monitoring features, the plurality of normal signal recording data and the plurality of abnormal signal recording data are used for training to obtain a signal anomaly detection model.

6. The method for detecting abnormality of live broadcast signal of an e-commerce platform according to claim 5, characterized in that: The preset quality rules include: If the bandwidth, delay, or packet loss rate exceeds the corresponding bandwidth threshold, delay threshold, or packet loss rate threshold, or the frame rate is lower than the normal frame rate range, or the audio and video synchronization is lower than the preset threshold and the video screen is stuck, it is judged as abnormal signal recording data.

7. The method for detecting abnormality of live broadcast signal of an e-commerce platform according to claim 1, characterized in that: The method further comprises: Screening abnormal signal data in the signal detection judgment data to obtain an abnormal probability value output by the signal anomaly detection model; Determining quality characteristics corresponding to the abnormal signal data; Compare the current abnormal signal with the historical abnormal signal and calculate the abnormal signal matching degree; Performing confidence scoring on the abnormal signal data, and classifying the abnormal signal according to the confidence scoring result; The confidence score formula is as follows: Among them, A is the confidence score, P model represents the abnormal probability value, F i represents the normalized value of the i-th feature of the abnormal signal, w i represents the weight of the i-th feature, S history represents the abnormal signal matching degree, α, β, and γ represent weight parameters.

8. An e-commerce platform live broadcast signal anomaly detection system, characterized in that: The system is used to implement the method for detecting abnormality of live broadcast signals of an e-commerce platform according to any one of claims 1 to 7, and the system comprises: A streaming media signal data acquisition module is used to connect to a target streaming media platform and acquire multiple streaming media signal data; A signal quality characteristic data extraction module, used to monitor the quality of the plurality of streaming media signal data and extract a plurality of signal quality characteristic data; A signal detection judgment data acquisition module, used for inputting the plurality of signal quality feature data into a signal anomaly detection model for judgment to obtain signal detection judgment data; The live signal detection report generation module is used to generate a live signal detection report for visual display based on the signal detection judgment data.

Images