Multistage control method based on Internet information security

Abstract

The invention discloses a multi-level control method based on Internet information security, and the method comprises the steps: constructing a community smart screen playing mode judgment model, and determining the playing modes of a community smart screen in a use peak period and a non-use peak period; the network module is used for cutting off the connection between the community smart screen and an external network in a non-use peak period, and obtaining a latest playlist of the community smart screen in a random networking mode; using a community smart screen playing mode judgment model to determine a playing mode of the community smart screen during holidays and festivals; determining an optimal playing mode in the current scene by using a community smart screen playing mode judgment model; and identifying abnormal activity information according to the network flow data and the access log, disconnecting the network connection of the community smart screen, and entering a physical isolation state. Through the multi-level control method, accurate content playing, dynamic adjustment and efficient safety management of the community smart screen are achieved, and the network safety and operation stability of the community smart screen are effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of Internet information security, and in particular to a multi-level control method based on Internet information security. Background Art

[0002] In the intelligent construction of modern cities, community smart screens, as the main carrier of information dissemination and interaction, have become an important part of community management and public services. However, with the increasing application of smart screens, they face a series of technical problems in actual operation, which directly affect the effectiveness, user experience and security of the system. First, the accuracy of the content played by community smart screens and the user experience are becoming increasingly prominent. The traditional playback mode usually relies on preset fixed content, which is difficult to dynamically respond to changes in user needs and actual situations. This inflexible playback mechanism often leads to the inability of the content played in different time periods to effectively meet the actual needs of current users, especially during peak and non-peak hours. As a result, users may receive content that is irrelevant or unrelated to their needs, which in turn affects their interactive experience and satisfaction with the system. In addition, there are significant differences between user behavior and population distribution during holidays and usual times, but it is difficult for the existing system to effectively adjust the playback content according to the changes in these special periods, further exacerbating the lack of user experience. Secondly, the network security problem of smart screens is becoming more and more serious. Community smart screens usually need to be connected to the Internet for a long time to update the playback content, but this continuous online state makes them easy targets for cyber attacks. Especially during non-peak usage periods, the network demand of smart screens is reduced, but the system remains connected to the Internet, increasing the time and probability of exposure to external threats. Current security protection measures mainly rely on traditional means such as firewalls and antivirus software. However, these measures are often passive defenses and it is difficult to respond quickly when a network attack occurs. The lack of the ability to quickly detect and isolate abnormal network activities may cause the community smart screen system to be paralyzed when it encounters an attack, resulting in the risk of information dissemination interruption and user data leakage. In addition, the balance between the timeliness of playlist updates and network security is becoming increasingly prominent. With the increase in user demand for real-time content, frequent updates of playlists have become necessary, but this is also accompanied by an increase in network security risks. Frequent online updates of content increase the chance of being attacked, while untimely content updates may cause the playback content to be outdated and unable to meet the real-time needs of users. This contradiction between timeliness and security is often difficult to effectively resolve in existing systems. The lack of a scientific evaluation and adjustment mechanism makes it difficult for the system to reduce network security risks while ensuring the timeliness of playback content. Therefore, the current community smart screen system has obvious problems in the accuracy of playback content, user experience, network security, and the balance between timeliness and security of playlist updates. These problems not only limit the system's operating efficiency and user satisfaction, but also pose a threat to the overall security of the system. Solving these technical problems is the key to improving the reliability and security of community smart screen systems. Summary of the invention

[0003] The present invention solves the problems existing in the above-mentioned prior art and provides a multi-level control method based on Internet information security, which mainly includes:

[0004] Through the historical scenario monitoring data of community smart screens, the peak usage hours of community smart screens are identified, and a community smart screen playback mode judgment model is constructed to determine the playback mode during the peak usage hours and non-peak usage hours of community smart screens, and update the playlist of community smart screens;

[0005] Through the network module, the community smart screen is disconnected from the external network during non-peak hours, and the latest community smart screen playlist is obtained from the cloud server by random networking and saved to the local server;

[0006] Based on the historical interaction data and historical crowd distribution data during the peak usage hours and non-peak usage hours during each holiday, the community smart screen playback mode judgment model is used to determine the playback mode of the community smart screen during the peak usage hours and non-peak usage hours during each holiday;

[0007] Based on the video stream and image data, the crowd flow in the current area is identified, and the community smart screen playback mode judgment model is used to determine the best playback mode for the current scenario, and the best playback content for the current scenario is updated through network transmission;

[0008] Acquire network traffic data and access logs in real time, identify abnormal activity information, and disconnect the community smart screen’s network connection through the network control module to enter a physical isolation state.

[0009] Furthermore, the historical scenario monitoring data of the community smart screen is used to identify the peak usage period of the community smart screen, and a community smart screen playback mode judgment model is constructed to determine the playback mode of the community smart screen during the peak usage period and the non-peak usage period, and to update the playlist of the community smart screen, including:

[0010] Through the historical usage records of community smart screens, historical scenario monitoring data of community smart screens are obtained and organized into the format of time series data. The scenario monitoring data includes crowd flow data and user interaction data; the ARIMA model is used to process the obtained time series data, extract peaks and trends in the data, and identify the peak usage periods of community smart screens; through the historical usage records of community smart screens, historical interaction data and historical crowd distribution data of community smart screens during peak usage periods and non-peak usage periods are obtained. The interaction data includes interaction frequency and interaction content, and the crowd distribution data includes crowd residence time, crowd density and crowd age distribution; based on the historical interaction data and historical crowd distribution data of community smart screens during peak usage periods and non-peak usage periods, the random forest algorithm is used for model training to build a community smart screen playback mode judgment model to determine the playback mode of the community smart screen during peak usage periods and non-peak usage periods. The playback mode includes playback content type and playback order; the playback content corresponding to the playback content type during peak usage periods and non-peak usage periods is obtained through the cloud server, the playback management module is used to adjust the playback content and order, the play list of the community smart screen is updated, and the play list of the community smart screen during non-peak usage periods is stored on the local server.

[0011] Furthermore, the network module cuts off the connection between the community smart screen and the external network during non-peak hours, and uses a random networking method to obtain the latest playlist of the community smart screen from the cloud server and save it to the local server, including:

[0012] Adopt a preset offline plan, control the offline operation of the community smart screen during non-peak hours through the network module, and cut off the connection with the external network; obtain the local server connection information of the community smart screen in the offline operation state, use the automatic connection module to configure the connection parameters, establish a connection between the community smart screen in the offline state and the local server, and obtain the playlist of the community smart screen in the corresponding time period through the local server. The local server connection information includes the IP address and port number of the local server; use a random number generator to generate the time point and networking method for network transmission update, and start the network transmission at a randomly generated time point through the network module, obtain the latest community smart screen playlist from the cloud server and save it to the local server, and cut off the connection with the external network again after saving. The networking methods include wifi networking, sim card networking and Ethernet networking;

[0013] It also includes evaluating the timeliness and network security of playlist updates based on the playlist update cycle and user interaction data of the community smart screen, and adjusting the playlist of the community smart screen by calculating the playlist weight coefficient;

[0014] The playlist update cycle and user interaction data based on the community smart screen evaluate the timeliness and network security of the playlist update, and adjust the playlist of the community smart screen by calculating the playlist trade-off coefficient, specifically including:

[0015] Through the historical usage records of the community smart screen, obtain the playlist update cycle and user interaction data of the community smart screen, and use the playlist update timeliness formula Calculate the timeliness of the playlist update of the community smart screen after the i-th network transmission, where M is the number of times the user accesses the content, and C n (t i ) is the time t j The vector representation of the playback mode of the latest playlist received by the community smart screen at that time, C o (t i ) is the time t j The vector representation of the play mode of the playlist before the community smart screen receives the latest playlist. The vector elements include the play content type and play order. i The larger the value, the more timely the content update and the better the timeliness. Through the historical network security record documents, the historical time point data of the community smart screen being attacked is obtained, and the ARIMA model is used to predict the time point when the community smart screen will be attacked next time. According to the prediction error formula Among them, t k is the actual time point of networking. is the predicted time point when the community smart screen will be attacked next time, n is the number of samples; through the network security index formula Calculate the security index of the ith network transmission, where H i is the entropy of the time point, indicating the random distribution of the time point, E i The error in predicting the time of Internet connection for the attacker, is the average value of the prediction error, used as a benchmark; H i The calculation formula is Among them, p k is the frequency of online time points in time period k, K is the total number of time periods; according to the timeliness of playlist updates and network security, the playlist weight coefficient formula is used Calculate the playlist trade-off coefficient, where S is the trade-off coefficient between security and timeliness of network transmission, N is the set number of network transmissions, and Δt i is the time interval between the ith network transmission, σ i The standard deviation of the time distribution of the i-th networking is obtained by obtaining the initial random number generator parameter P0, defining the range and distribution pattern of the generated random time points, and determining Δt i and σ i; If the playlist weight coefficient S exceeds the preset fluctuation range, the playlist of the community smart screen is adjusted until the playlist weight coefficient S is within the preset fluctuation range; for each generated networking time point, the corresponding playlist weight coefficient S is calculated, and the content update timeliness T after each networking is recorded i and safety index R i , dynamically adjust the random number generation strategy.

[0016] Furthermore, the community smart screen playback mode judgment model is used to determine the playback mode of the community smart screen during the peak usage period and the non-peak usage period during each holiday according to the historical interaction data and historical crowd distribution data during the peak usage period and the non-peak usage period during each holiday, including:

[0017] Through the historical usage records of community smart screens, the historical interaction data and historical crowd distribution data of the peak usage periods and non-peak usage periods during each holiday are obtained, and the peak usage periods and non-peak usage periods during each holiday are determined using the ARIMA model; through the historical usage records of community smart screens, the historical interaction data and historical crowd distribution data of the peak usage periods and non-peak usage periods of the community smart screens during each holiday are obtained, and the community smart screen playback mode judgment model is used to determine the playback mode of the peak usage periods and non-peak usage periods of the community smart screens during each holiday; the playlists of the community smart screens during the peak usage periods and non-peak usage periods of the community smart screens during each holiday are obtained through the cloud server, and before the community smart screens cut off the connection with the external network during the holiday, the playlists of the community smart screens during the non-peak usage periods during the holiday are stored on the local server.

[0018] Furthermore, the crowd flow in the current area is identified based on the video stream and image data, the optimal playback mode in the current scenario is determined using the community smart screen playback mode judgment model, and the optimal playback content in the current scenario is updated through network transmission, including:

[0019] Through the video acquisition module of the community smart screen, video streams and image data are obtained in real time, and the OpenCV algorithm is used to analyze the video streams and image data, identify the crowd flow in the current area, and compare the preset crowd flow threshold to determine whether the crowd flow exceeds the preset crowd flow threshold; if the crowd flow exceeds the preset crowd flow threshold, real-time scenario monitoring data is obtained, and the community smart screen playback mode judgment model is used to determine the best playback mode in the current scenario; the current playback mode of the community smart screen is obtained through the play list of the community smart screen, and the best playback mode and the current playback mode in the current scenario are respectively converted into vector representations, and the vector elements include the playback content type and the playback order. The cosine similarity algorithm is used to calculate the similarity of the two vectors, and the calculated similarity is compared with the preset similarity threshold; if the similarity is lower than the preset similarity threshold, the best playback content of the current scenario is updated through network transmission, and the playback management module is used to adjust the playback content and order, and the play list of the community smart screen is updated.

[0020] Furthermore, the real-time acquisition of network traffic data and access logs, identification of abnormal activity information, and disconnection of the community smart screen's network connection through the network control module to enter a physical isolation state include:

[0021] Through the network monitoring module connected to the network interface of the community smart screen, network traffic data and access logs are obtained in real time; based on the network traffic data and access logs, the isolation forest algorithm is used to determine whether there are any abnormalities in the current network activity; based on the anomaly detection results, information for identifying abnormal activities is obtained, including the source IP address, access time and behavioral characteristics, and a decision tree algorithm is used for model training to determine whether the abnormal activity is a virus or hacker attack; the security management module is configured, and if the abnormal activity is determined to be a virus or hacker attack, the network connection of the community smart screen is disconnected through the network control module, the screen enters a physical isolation state, and the community smart screen is controlled to stop broadcasting; the operation status and operation log of the device are obtained according to the operation monitoring module of the community smart screen, and the security management module is used to determine whether the release conditions are met. If so, the physical isolation state is released, and the community smart screen is controlled to resume broadcasting.

[0022] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0023] The present invention provides a multi-level control method based on Internet information security. Through the analysis of historical scenario data, the present invention can intelligently identify peak and non-peak time periods, dynamically adjust the playback mode, make the playback content closer to user needs, and thus improve the user experience. At the same time, by cutting off the network connection during non-peak usage periods and using a random networking method to update the content, the network security risk is greatly reduced, and potential attacks and data leaks are avoided. The present invention also uses the playlist update cycle and user interaction data to scientifically evaluate and balance the timeliness and security of the playlist to ensure the timely update and security maintenance of the playback content. With the help of real-time analysis of video streams and image data, the playback content can be adjusted according to the actual scene to improve the emergency response capability. By real-time monitoring of network traffic and access logs, abnormal network activities can be quickly identified and isolated to ensure the security and stability of the community smart screen when encountering network threats. The present invention realizes accurate content playback, dynamic adjustment and efficient security management of the community smart screen through a multi-level control method, while improving the playback effect and user satisfaction of the community smart screen, effectively ensuring the network security and operation stability of the community smart screen. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flowchart of a multi-level control method based on Internet information security of the present invention;

[0025] Figure 2 A schematic diagram of a multi-level control method based on Internet information security of the present invention; DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] like Figure 1-2 In this embodiment, a multi-level control method based on Internet information security may specifically include:

[0028] Step S101, through the historical scenario monitoring data of the community smart screen, identify the peak usage period of the community smart screen, and build a community smart screen playback mode judgment model to determine the playback mode of the community smart screen during peak usage periods and non-peak usage periods, and update the play list of the community smart screen.

[0029] Through the historical usage records of community smart screens, the historical scenario monitoring data of community smart screens are obtained and organized into the format of time series data. The scenario monitoring data includes crowd flow data and user interaction data. The ARIMA model is used to process the obtained time series data, extract peaks and trends in the data, and identify the peak usage period of community smart screens. Through the historical usage records of community smart screens, the historical interaction data and historical crowd distribution data of the peak usage period and non-peak usage period of community smart screens are obtained. The interaction data includes interaction frequency and interaction content, and the crowd distribution data includes crowd residence time, crowd density and crowd age distribution. According to the historical interaction data and historical crowd distribution data of the peak usage period and non-peak usage period of community smart screens, the random forest algorithm is used for model training to build a community smart screen playback mode judgment model to determine the playback mode of the community smart screen during the peak usage period and non-peak usage period. The playback mode includes the playback content type and playback order. The playback content corresponding to the playback content type during the peak usage period and non-peak usage period is obtained through the cloud server, and the playback management module is used to adjust the playback content and order, update the playlist of the community smart screen, and store the playlist of the community smart screen during the non-peak usage period to the local server.

[0030] Exemplarily, through the historical usage records of community smart screens, the historical scenario monitoring data of community smart screens in the past month are obtained. The data records the daily crowd flow and user interaction data, among which the crowd flow reaches a peak between 8:00 and 10:00 in the morning and between 5:00 and 7:00 in the afternoon. At the same time, the user interaction data also shows high-frequency interactive operations during these periods, such as clicks and information queries. These data are transmitted to the edge computer, where the data are sorted into a time series format, and the ARIMA model is applied to analyze the data, identifying that the peak usage hours are mainly concentrated in the morning from 8:00 to 10:00 and the afternoon from 17:00 to 19:00, while the non-peak usage hours are from 19:00 to 8:00 in the evening and from 10:00 in the morning to 17:00 in the afternoon. According to the interactive data of the community smart screen during these peak and non-peak hours, it is identified that the interaction frequency during peak hours is higher, and the interactive content is mainly concentrated in news browsing and traffic information query, while the interaction frequency during non-peak hours is lower, mainly community activity information and advertising display. At the same time, the crowd distribution data shows that during peak hours, the crowd stays for a short time, the density is high, and the age distribution tends to be young and middle-aged, while during non-peak hours, the crowd stays for a long time, the density is low, and the elderly are the main ones. Based on these data, the random forest algorithm was used to train the model of the data during peak and non-peak hours, and the playback mode judgment model of the community smart screen was constructed. The playback mode judgment model of the community smart screen determined that the playback mode during peak hours should be mainly short-term and high-frequency information content, such as news flashes, traffic dynamics, and emergency notifications. The playback content types are mainly news and traffic, and the playback order is prioritized for instant information. During non-peak hours, the playback mode should be mainly long-term and low-frequency content, such as community announcements, health lectures, and advertisements. The playback content types are mainly community service information and advertisements, and the playback order is prioritized for important announcements. The specific playback content corresponding to these playback modes was obtained through the cloud server. For example, the content types played during peak hours include the news summary of the day and real-time traffic information, and the content played during non-peak hours includes community event previews and health tips. The playback management module adjusts the content and order of the playlist based on these contents, and stores the adjusted playlist on the local server to ensure that the playback content is updated and the playback order is reasonable when the system is offline.

[0031] Step S102: During non-peak hours, the network module cuts off the connection between the community smart screen and the external network, and uses random networking to obtain the latest playlist of the community smart screen from the cloud server and save it to the local server.

[0032] Adopt a preset offline plan, and control the offline operation of the community smart screen during non-peak hours through the network module to cut off the connection with the external network. Obtain the local server connection information of the community smart screen in the offline operation state, use the automatic connection module to configure the connection parameters, establish a connection between the community smart screen in the offline state and the local server, and obtain the playlist of the community smart screen in the corresponding time period through the local server. The local server connection information includes the IP address and port number of the local server. Use a random number generator to generate the time point and networking method for network transmission updates, and start the network transmission at a randomly generated time point through the network module, obtain the latest community smart screen playlist from the cloud server and save it to the local server. After saving, cut off the connection with the external network again. The networking methods include wifi networking, sim card networking and Ethernet networking.

[0033] Exemplarily, in the operation of a community smart screen, according to the preset offline plan, during non-peak usage hours, such as 2:00 to 4:00 p.m., the connection between the community smart screen and the external network is cut off through the network module. The local server connection information of the community smart screen in the offline operation state is obtained, where the IP address of the local server is "192.168.1.100" and the port number is "8080". These connection parameters are configured using the automatic connection module, and the community smart screen is successfully connected to the local server, and the playlist for the period from 2:00 to 4:00 p.m. is obtained from the local server. The playlist includes community announcements and advertising content, with announcements first and advertisements second. In order to ensure the continuous updating of the playlist, a random number generator is used in the edge computer to generate two time points for network transmission updates, at 2:30 and 3:15 p.m., and two networking methods, Ethernet and Wi-Fi, are randomly selected. At 2:30 p.m., the network transmission was started via Ethernet, the latest community announcement content was obtained from the cloud server and saved to the local server, and at 3:15 p.m., the network was connected again via Wi-Fi to update the advertising content in the playlist. After each network transmission was completed, the connection with the external network was immediately cut off to ensure that the community smart screen remained offline for most of the time, thereby reducing network security risks and ensuring the timeliness and continuous updating of the playback content.

[0034] Among them, based on the playlist update cycle and user interaction data of the community smart screen, the timeliness of the playlist update and the network security are evaluated, and the playlist of the community smart screen is adjusted by calculating the playlist weight coefficient.

[0035] Through the historical usage records of the community smart screen, obtain the playlist update cycle and user interaction data of the community smart screen, and use the playlist update timeliness formula Calculate the timeliness of the playlist update of the community smart screen after the i-th network transmission, where M is the number of times the user accesses the content, and C n (t i ) is the time t j The vector representation of the playback mode of the latest playlist received by the community smart screen at that time, C o (t i ) is the time t j The vector representation of the play mode of the playlist before the community smart screen receives the latest playlist. The vector elements include the play content type and play order. i The larger the value, the more timely the content update and the better the timeliness. Through the historical network security record documents, the historical time point data of the community smart screen being attacked is obtained, and the ARIMA model is used to predict the time point when the community smart screen will be attacked next time. According to the prediction error formula Among them, t k is the actual time point of networking. is the predicted time point when the community smart screen will be attacked next time, and n is the number of samples. Calculate the security index of the ith network transmission, where H i is the entropy of the time point, indicating the random distribution of the time point, E i The error in predicting the time of Internet connection for the attacker, is the average value of the prediction error, used as a benchmark. i The calculation formula is Among them, p k is the frequency of online time points in time period k, and K is the total number of time periods. According to the timeliness of playlist updates and network security, the playlist weight coefficient formula is used. Calculate the playlist trade-off coefficient, where S is the trade-off coefficient between security and timeliness of network transmission, N is the set number of network transmissions, and Δt i is the time interval between the ith network transmission, σ i The standard deviation of the time distribution of the i-th networking is obtained by obtaining the initial random number generator parameter P0, defining the range and distribution pattern of the generated random time points, and determining Δt i and σ i If the playlist weight coefficient S exceeds the preset fluctuation range, the playlist of the community smart screen is adjusted until the playlist weight coefficient S is within the preset fluctuation range. For each generated networking time point, calculate the corresponding playlist weight coefficient S, and record the content update timeliness T after each networking. i and safety index R i , dynamically adjust the random number generation strategy.

[0036] For example, in the usage record of a community smart screen, it is detected that the number of times M that users accessed the playback content in the past day is 50 times, and the latest playlist update occurred at 10:00 am. The updated playback mode vector C n (t j )=[2,3,1], indicating that content types B, C, and A are played in sequence, and the play mode vector C before the update o (t j )=[2,1,3], indicating that content types B, A, and C are played in order, using the playlist update timeliness formula Calculate T i =0.4, where M is the number of times the user accesses the content, C n (t i ) is the time t j The vector representation of the playback mode of the latest playlist received by the community smart screen at that time, C o (t i ) is the time t j The vector representation of the play mode of the playlist before the community smart screen receives the latest playlist. The vector elements include the play content type and play order. i The larger the value, the more timely the content update and the better the timeliness. i =0.4 indicates that the timeliness of the content update is low. By analyzing the historical records of attacks on community smart screens, the system uses the ARIMA model to predict that the next possible attack time point is 2:00 pm, while the actual connection time is 11:00 am. According to the prediction error formula The calculated error is 0.5, where t k is the actual time point of networking. is the predicted time point when the community smart screen will be attacked next time, and n is the number of samples. Calculate the security index R of this network i , where H i is the entropy of the time point, indicating the random distribution of the time point, E i The error in predicting the time of Internet connection for the attacker, is the average value of the prediction error, which is used as a benchmark. Based on the historical network connection time frequency data of the community smart screen, the formula is used Calculate H i =0.7, therefore, R i =1.575, where p k is the frequency of network time points in time period k, K is the total number of time periods, R i=1.575 is greater than the preset entropy threshold of 1.2, indicating that the security of this network connection is relatively high. Use the playlist weight coefficient formula The calculation results show that S = 0.4725, where S is the trade-off coefficient between security and timeliness of network transmission, N is the set number of network transmissions, and Δt i is the time interval between the ith network transmission, σ i The standard deviation of the time distribution of the i-th networking is obtained by obtaining the initial random number generator parameter P0, defining the range and distribution pattern of the generated random time points, and determining Δt i and σ i At this time, S=0.4725 exceeds the preset fluctuation range of 0.5-0.7. Therefore, it is necessary to automatically adjust the playlist of the community smart screen until S returns to the preset range to ensure the balance between system security and timeliness.

[0037] Step S103, based on the historical interaction data and historical crowd distribution data during the peak usage periods and non-peak usage periods during each holiday, use the community smart screen playback mode judgment model to determine the playback mode during the peak usage periods and non-peak usage periods of the community smart screen during each holiday.

[0038] Through the historical usage records of community smart screens, obtain the historical interaction data and historical crowd distribution data of the peak usage hours and non-peak usage hours during each holiday, and use the ARIMA model to determine the peak usage hours and non-peak usage hours during each holiday. Through the historical usage records of community smart screens, obtain the historical interaction data and historical crowd distribution data of the peak usage hours and non-peak usage hours of community smart screens during each holiday, and use the community smart screen playback mode judgment model to determine the playback mode of the peak usage hours and non-peak usage hours of community smart screens during each holiday. Obtain the playlist of the community smart screen during the peak usage hours and non-peak usage hours of the community smart screen during each holiday through the cloud server, and before the community smart screen cuts off the connection with the external network during the holiday, store the playlist of the community smart screen during the non-peak usage hours during the holiday to the local server.

[0039] For example, through the historical usage records of the community smart screen, the usage records during the Spring Festival in the past three years were obtained, and the historical interaction data and crowd distribution data during the peak usage period and non-peak usage period during the period were obtained. These data are transmitted to the edge computer, where the ARIMA model is used for analysis to identify that the peak usage period during the Spring Festival is concentrated from 10:00 am to 12:00 noon and 5:00 pm to 8:00 pm every day, while the non-peak usage period is mainly from 2:00 pm to 4:00 pm and after 9:00 pm. During the peak period during the Spring Festival, the residents' interaction frequency is high, mainly focusing on sending holiday greetings and watching Spring Festival celebration videos, and the crowd stays for a short time, the density is high, and the age distribution is mainly young and middle-aged. During the non-peak period, the interaction frequency is significantly reduced, mainly to check community service information and activity arrangements, the stay time is long, the crowd density is low, and the proportion of the elderly is large. Based on these data, the playback mode judgment model of the community smart screen is used to determine the playback mode during the Spring Festival. The playback mode during peak hours includes playing holiday celebration videos, holiday greetings, and community event previews. The order of playback content is holiday celebration videos first, followed by greeting messages. During non-peak hours, the playback mode focuses on community service information, health lectures, and advertisements, with community service information first. Before the Spring Festival, the playback content corresponding to these playback modes was obtained through the cloud server, and the playlists during peak hours were directly updated to the community smart screen. At the same time, before the community smart screen cuts off the connection to the external network, the playlists during non-peak hours are stored on the local server to ensure that the community smart screen can still play content according to the predetermined playback mode when it is offline during the Spring Festival. In this way, even during non-peak usage periods during the Spring Festival, the community smart screen can continue to play relevant content according to the playlists stored on the local server, ensuring that residents enjoy continuous services and rich content during the holidays.

[0040] Step S104, based on the video stream and image data, identify the crowd flow in the current area, use the community smart screen playback mode judgment model to determine the best playback mode for the current scenario, and update the best playback content for the current scenario through network transmission.

[0041] Through the video acquisition module of the community smart screen, video streams and image data are obtained in real time, and the OpenCV algorithm is used to analyze the video streams and image data, identify the crowd flow in the current area, compare the preset crowd flow threshold, and determine whether the crowd flow exceeds the preset crowd flow threshold. If the crowd flow exceeds the preset crowd flow threshold, real-time scenario monitoring data is obtained, and the community smart screen playback mode judgment model is used to determine the best playback mode for the current scenario. The current playback mode of the community smart screen is obtained through the playlist of the community smart screen, and the best playback mode and the current playback mode in the current scenario are converted into vector representations respectively. The vector elements include the playback content type and the playback order. The cosine similarity algorithm is used to calculate the similarity of the two vectors, and the calculated similarity is compared with the preset similarity threshold. If the similarity is lower than the preset similarity threshold, the best playback content for the current scenario is updated through network transmission, and the playback management module is used to adjust the playback content and order, and the playlist of the community smart screen is updated.

[0042] For example, the video acquisition module of a community smart screen acquires the video stream and image data of the square at 10:30 in the morning in real time, transmits the data to the edge computer, uses the OpenCV algorithm to analyze in the edge computer, and calculates that the crowd flow in the current square area is 120 people, while the preset crowd flow threshold is 100 people. Since the current crowd flow exceeds the preset threshold, the real-time scenario monitoring data is further acquired, and the community smart screen playback mode judgment model is used to determine that the best playback mode in the current scenario is to play emergency announcements and crowd diversion information, and the order of playback content is emergency announcements first. The current playback mode is obtained through the playlist of the community smart screen, and it is found that the current playback content is playing community activity promotional videos, and the playback order is promotional videos first. The best playback mode and the existing playback mode in the current scenario are converted into vector representations respectively: the best playback mode vector is [emergency announcement, crowd diversion], and the current playback mode vector is [community activity promotional video]. Using the cosine similarity algorithm, the similarity of the two vectors is calculated to be 0.2, and the preset similarity threshold is 0.8. Since the calculated similarity is far below the preset threshold, the best playback content for the current scenario is immediately updated through network transmission, the latest emergency announcements and crowd control information are obtained from the cloud, and the playback management module is used to adjust the playback content and order. The updated playlist is set to prioritize emergency announcements to ensure that people can receive important information in a timely manner and take corresponding actions.

[0043] Step S105, obtain network traffic data and access logs in real time, identify abnormal activity information, and disconnect the network connection of the community smart screen through the network control module to enter a physical isolation state.

[0044] Through the network monitoring module connected to the network interface of the community smart screen, network traffic data and access logs are obtained in real time. Based on the network traffic data and access logs, the isolation forest algorithm is used to determine whether there are any abnormalities in the current network activity. According to the anomaly detection results, the information for identifying abnormal activities is obtained, including the source IP address, access time and behavioral characteristics, and the decision tree algorithm is used for model training to determine whether the abnormal activity belongs to a virus or hacker attack. Configure the security management module. If the abnormal activity is determined to be a virus or hacker attack, the network connection of the community smart screen is disconnected through the network control module, entering the physical isolation state, and controlling the community smart screen to stop broadcasting. According to the operation monitoring module of the community smart screen, the operation status and operation log of the device are obtained, and the security management module is used to determine whether the release conditions are met. If so, the physical isolation state is lifted, and the community smart screen is controlled to resume broadcasting.

[0045] For example, on an ordinary working day, the network monitoring module of the community smart screen obtains network traffic data and access logs in real time through the connection with the network interface. It was detected that the network traffic had an abnormal peak at 11:45 am, and the transmission speed and number of data packets were significantly higher than the usual level. The isolation forest algorithm is used to analyze these network activities and determine that there is abnormal behavior in the current network. According to the detection results, an abnormal activity was identified. The detailed information shows that the source IP address of the activity is "192.168.1.15", the access time is 11:45, and the behavioral characteristics show that the IP address has requested access to the unauthorized community smart screen management page multiple times in a short period of time. The decision tree algorithm is used to train and judge the model of abnormal activities. The results show that this activity is very likely to be a malicious hacker attack. Based on this judgment, the security management module is automatically configured, and the community smart screen is immediately disconnected from the external network through the network control module, and the device is switched to a physically isolated state. At the same time, the community smart screen is controlled to stop all ongoing playback content to prevent the attack from spreading further or causing damage. In the following period, the operation monitoring module of the community smart screen continuously obtains the operation status and operation logs of the device, and evaluates the safety status of the device by analyzing these data. After a period of monitoring and analysis, it is determined that all safety conditions of the device have been met, and then the physical isolation status is lifted through the security management module, the network is reconnected, and the normal broadcast of the community smart screen is restored to ensure that community services can continue smoothly.

[0046] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the concept of the present application. The above exemplary features are replaced with the technical features with similar functions disclosed in the present application (but not limited to) to form a technical solution.

Claims

1. A multi-level control method based on Internet information security, characterized in that: The method comprises: Through the historical scenario monitoring data of community smart screens, the peak usage hours of community smart screens are identified, and a community smart screen playback mode judgment model is constructed to determine the playback mode during the peak usage hours and non-peak usage hours of community smart screens, and update the playlist of community smart screens; Through the network module, the community smart screen is disconnected from the external network during non-peak hours, and the latest community smart screen playlist is obtained from the cloud server by random networking and saved to the local server; Based on the historical interaction data and historical crowd distribution data during the peak usage hours and non-peak usage hours during each holiday, the community smart screen playback mode judgment model is used to determine the playback mode of the community smart screen during the peak usage hours and non-peak usage hours during each holiday; Based on the video stream and image data, the crowd flow in the current area is identified, and the community smart screen playback mode judgment model is used to determine the best playback mode for the current scenario, and the best playback content for the current scenario is updated through network transmission; Acquire network traffic data and access logs in real time, identify abnormal activity information, and disconnect the community smart screen’s network connection through the network control module to enter a physical isolation state.

2. The method according to claim 1, wherein: The method of identifying the peak usage period of the community smart screen through the historical scenario monitoring data of the community smart screen, building a community smart screen playback mode judgment model, determining the playback mode of the community smart screen during the peak usage period and the non-peak usage period, and updating the playlist of the community smart screen includes: Through the historical usage records of community smart screens, historical scenario monitoring data of community smart screens are obtained and organized into the format of time series data. The scenario monitoring data includes crowd flow data and user interaction data; the ARIMA model is used to process the obtained time series data, extract peaks and trends in the data, and identify the peak usage periods of community smart screens; through the historical usage records of community smart screens, historical interaction data and historical crowd distribution data of community smart screens during peak usage periods and non-peak usage periods are obtained. The interaction data includes interaction frequency and interaction content, and the crowd distribution data includes crowd residence time, crowd density and crowd age distribution; based on the historical interaction data and historical crowd distribution data of community smart screens during peak usage periods and non-peak usage periods, the random forest algorithm is used for model training to build a community smart screen playback mode judgment model to determine the playback mode of the community smart screen during peak usage periods and non-peak usage periods. The playback mode includes playback content type and playback order; the playback content corresponding to the playback content type during peak usage periods and non-peak usage periods is obtained through the cloud server, the playback management module is used to adjust the playback content and order, the play list of the community smart screen is updated, and the play list of the community smart screen during non-peak usage periods is stored on the local server.

3. The method according to claim 1, wherein: The network module cuts off the connection between the community smart screen and the external network during non-peak hours, and uses a random networking method to obtain the latest playlist of the community smart screen from the cloud server and save it to the local server, including: Adopt a preset offline plan, control the community smart screen to operate offline during non-peak hours through the network module, and cut off the connection with the external network; Obtain the local server connection information of the community smart screen in the offline running state, use the automatic connection module to configure the connection parameters, establish a connection between the community smart screen in the offline state and the local server, and obtain the playlist of the community smart screen in the corresponding time period through the local server. The local server connection information includes the IP address and port number of the local server; use a random number generator to generate the time point and networking method for network transmission update, and start the network transmission at the randomly generated time point through the network module, obtain the latest community smart screen playlist from the cloud server and save it to the local server. After saving, cut off the connection with the external network again. The networking methods include wifi networking, sim card networking and Ethernet networking; Based on the playlist update cycle and user interaction data of the community smart screen, the timeliness and network security of the playlist update are evaluated, and the playlist of the community smart screen is adjusted by calculating the playlist weight coefficient.

4. The method according to claim 3, wherein: The playlist update cycle and user interaction data based on the community smart screen evaluate the timeliness and network security of the playlist update, and adjust the playlist of the community smart screen by calculating the playlist weight coefficient, including: Through the historical usage records of the community smart screen, obtain the playlist update cycle and user interaction data of the community smart screen, and use the playlist update timeliness formula Calculate the timeliness of the playlist update of the community smart screen after the i-th network transmission, where M is the number of times the user accesses the content, and C n (t i ) is the time t j The vector representation of the playback mode of the latest playlist received by the community smart screen at that time, C o (t i ) is the time t j The vector representation of the play mode of the playlist before the community smart screen receives the latest playlist. The vector elements include the play content type and play order. i The larger the value, the more timely the content update and the better the timeliness. Through the historical network security record documents, the historical time point data of the community smart screen being attacked is obtained, and the ARIMA model is used to predict the time point when the community smart screen will be attacked next time. According to the prediction error formula Among them, t k is the actual time point of networking. is the predicted time point when the community smart screen will be attacked next time, n is the number of samples; through the network security index formula Calculate the security index of the ith network transmission, where H i is the entropy of the time point, indicating the random distribution of the time point, E i The error in predicting the time of Internet connection for the attacker, is the average value of the prediction error, used as a benchmark; H i The calculation formula is Among them, p k is the frequency of online time points in time period k, K is the total number of time periods; according to the timeliness of playlist updates and network security, the playlist weight coefficient formula is used Calculate the playlist trade-off coefficient, where S is the trade-off coefficient between security and timeliness of network transmission, N is the set number of network transmissions, and Δt i is the time interval between the ith network transmission, σ i The standard deviation of the time distribution of the i-th networking is obtained by obtaining the initial random number generator parameter P0, defining the range and distribution pattern of the generated random time points, and determining Δt i and σ i ; If the playlist weight coefficient S exceeds the preset fluctuation range, the playlist of the community smart screen is adjusted until the playlist weight coefficient S is within the preset fluctuation range; for each generated networking time point, the corresponding playlist weight coefficient S is calculated, and the content update timeliness T after each networking is recorded i and safety index R i , dynamically adjust the random number generation strategy.

5. The method according to claim 1, wherein: The community smart screen playback mode judgment model is used according to the historical interaction data and historical crowd distribution data during the peak usage period and the non-peak usage period during each holiday to determine the playback mode of the community smart screen during the peak usage period and the non-peak usage period during each holiday, including: Through the historical usage records of community smart screens, the historical interaction data and historical crowd distribution data of the peak usage periods and non-peak usage periods during each holiday are obtained, and the peak usage periods and non-peak usage periods during each holiday are determined using the ARIMA model; through the historical usage records of community smart screens, the historical interaction data and historical crowd distribution data of the peak usage periods and non-peak usage periods of the community smart screens during each holiday are obtained, and the community smart screen playback mode judgment model is used to determine the playback mode of the peak usage periods and non-peak usage periods of the community smart screens during each holiday; the playlists of the community smart screens during the peak usage periods and non-peak usage periods of the community smart screens during each holiday are obtained through the cloud server, and before the community smart screens cut off the connection with the external network during the holiday, the playlists of the community smart screens during the non-peak usage periods during the holiday are stored on the local server.

6. The method according to claim 1, wherein: The method of identifying the crowd flow in the current area based on the video stream and image data, using the community smart screen playback mode judgment model to determine the best playback mode in the current scenario, and updating the best playback content in the current scenario through network transmission includes: Through the video acquisition module of the community smart screen, video streams and image data are obtained in real time, and the OpenCV algorithm is used to analyze the video streams and image data, identify the crowd flow in the current area, and compare the preset crowd flow threshold to determine whether the crowd flow exceeds the preset crowd flow threshold; if the crowd flow exceeds the preset crowd flow threshold, obtain real-time scenario monitoring data, use the community smart screen playback mode judgment model to determine the best playback mode in the current scenario; The current playback mode of the community smart screen is obtained through the playlist of the community smart screen, and the optimal playback mode and the current playback mode in the current scenario are converted into vector representations respectively. The vector elements include the playback content type and the playback order. The cosine similarity algorithm is used to calculate the similarity of the two vectors, and the calculated similarity is compared with the preset similarity threshold; If the similarity is lower than the preset similarity threshold, the best playback content for the current scenario is updated through network transmission, the playback content and order are adjusted using the playback management module, and the play list of the community smart screen is updated.

7. The method according to claim 1, wherein: The real-time acquisition of network traffic data and access logs, identification of abnormal activity information, and disconnection of the community smart screen's network connection through the network control module to enter a physical isolation state include: Through the network monitoring module connected to the network interface of the community smart screen, network traffic data and access logs are obtained in real time; based on the network traffic data and access logs, the isolation forest algorithm is used to determine whether there are any abnormalities in the current network activity; According to the results of anomaly detection, obtain information on identifying abnormal activities, including the source IP address, access time, and behavioral characteristics, and use the decision tree algorithm to train the model to determine whether the abnormal activity is a virus or hacker attack; configure the security management module. If it is determined that the abnormal activity is a virus or hacker attack, the network connection of the community smart screen is disconnected through the network control module, entering a physical isolation state, and controlling the community smart screen to stop broadcasting; obtain the operating status and operation log of the device according to the operation monitoring module of the community smart screen, and determine whether the release conditions are met through the security management module. If so, the physical isolation state is released, and the community smart screen is controlled to resume broadcasting.

Images