AI-based information transmission resource intelligent recommendation and automatic scheduling method and system

Abstract

The application relates to an AI-based information transmission resource intelligent recommendation and automatic scheduling method under safe interaction, which comprises the following steps: inputting terminal information and content attribute data; AI analyzing the terminal information and the content attribute data, and generating a scheduling instruction based on the analysis result; directly mapping the scheduling instruction into a pixel channel value; converting the pixel channel value into a pixel straight-through bearing signal and outputting the pixel straight-through bearing signal to a one-way transmission link, which one-way transmits the pixel straight-through bearing signal to an inner network terminal control and execution module; decoding, re-timing and restoring the signal into a byte stream by the inner network terminal control and execution module; recombining the byte stream into an original scheduling instruction based on a preset video display format; and executing the scheduling instruction to control the playing of each terminal one-way connected with the terminal according to the scheduling instruction. Through the application, automatic decision can be made based on real-time terminal state, optimal coupling of "content" and "terminal" is realized, and the safety of the inner network is ensured.

Description

Technical Field

[0001] This invention relates to the field of smart city information dissemination and communication resource scheduling technology, and in particular to a method and system for intelligent recommendation and automatic scheduling of information transmission resources based on big data collection and AI (artificial intelligence) that enables secure interaction between the external network and the internal network. Background Technology

[0002] With the deepening of urban informatization, information publishing systems have been widely applied in many fields such as urban governance, public safety, commercial display, and building management. Traditional information publishing systems rely on manual scheduling, resulting in low content delivery efficiency. Moreover, in public safety or sensitive government intranet publishing environments, information transmission involved in external interactions usually needs to pass through a "cross-domain one-way channel" to ensure network security. Traditional network transmission protocols often face problems such as complex packet encapsulation, difficulty in retransmission verification, and poor real-time performance on one-way links. Summary of the Invention

[0003] To address the above technical problems, this application proposes an AI-based intelligent recommendation and automatic scheduling method for information transmission resources, comprising the following steps:

[0004] Input the terminal information and content attribute data;

[0005] AI is used to analyze information and content attribute data from various terminals, and scheduling instructions are generated based on the analysis results.

[0006] Based on a fixed preset video display format, scheduling instructions are directly mapped to pixel channel values;

[0007] The pixel channel value is converted into a pixel pass-through bearer signal and output to a unidirectional transmission link;

[0008] The unidirectional transmission link transmits the pixel pass-through bearer signal unidirectionally to the intranet terminal control and execution module;

[0009] The intranet terminal control and execution module decodes the received signal based on the preset video display format, re-times it, and restores the signal to a byte stream; it reassembles the byte stream into the original scheduling instructions based on the preset video display format; and it executes the scheduling instructions to control the playback of each terminal that is unidirectionally connected to it according to the scheduling instructions.

[0010] According to some embodiments of the present invention, the terminal information includes terminal feature data, historical delivery effect data, and geographical location and scene data; the scheduling instructions include: playback time, playback content, push priority, Top-N list of best playback terminals, best coverage range, best coverage terminal for emergency content, instruction execution order, and multi-terminal synchronization strategy.

[0011] According to some embodiments of the present invention, the step of analyzing terminal information and content attribute data through AI and generating scheduling instructions based on the analysis results includes: generating audience profile vectors based on terminal feature data and historical delivery effect data; generating content semantic vectors based on content attribute data; identifying geographical / scene hotspots; calculating recommendation scores based on audience profile vectors, content semantic vectors, and geographical / scene hotspots; and automatically generating the scheduling instructions based on the recommendation scores.

[0012] According to some embodiments of the present invention, the frame encapsulation structure of the scheduling instruction includes: a frame header, a frame body, a line tail, and a frame tail; the frame body contains a line body, which serves as the actual carrier of the scheduling instruction; the scheduling instruction is mapped in pixel channel order, and the intranet terminal control and execution module reads the pixel channel values ​​line by line, directly forms a byte stream according to RGB corresponding to 3 bytes, and restores the pixel channel values ​​to the byte stream.

[0013] According to some embodiments of the present invention, multiple scheduling instructions are concatenated and encapsulated into an image frame. The frame header of the image frame contains an index table, which records the identity sequence number of each scheduling instruction, as well as the starting offset and length of the scheduling instruction in the image frame. The frame body structure arranges the scheduling instructions sequentially. After an image frame is encapsulated, it is sent to a pixel bearer mapper or the encapsulated image frame is sent at predetermined time intervals.

[0014] This application also provides an AI-based intelligent recommendation and automatic scheduling system for information transmission resources, characterized in that it includes: a data input unit, an AI intelligent analysis and scheduling unit, and a scheduling execution unit;

[0015] The data input unit is used to input terminal information and content attribute data;

[0016] The AI ​​intelligent analysis and scheduling unit is connected to the data input unit. It analyzes the information and content attribute data of each terminal through AI and generates scheduling instructions based on the analysis results.

[0017] The scheduling execution unit is connected to the AI ​​intelligent analysis and scheduling unit and is used to execute the scheduling instructions;

[0018] The scheduling execution unit includes a scheduling instruction packager, a one-way transmission link, and an intranet terminal control and execution module;

[0019] The scheduling instruction packer includes a pixel bearer mapper and a video transmitter; the intranet terminal control and execution module includes a decoding and retiming module, a scheduling instruction reassembler, and an instruction execution module.

[0020] The pixel bearer mapper is used to receive the scheduling instruction and directly map the scheduling instruction into pixel channel values ​​based on a fixed preset video display format;

[0021] The video transmitter is connected to the pixel bearer mapper, which is used to convert the pixel channel values ​​into pixel pass-through bearer signals and output them to the unidirectional transmission link;

[0022] The unidirectional transmission link is used to transmit the pixel pass-through bearer signal unidirectionally to the decoding and retiming module;

[0023] The decoding and retiming module is used to decode the received signal, retime it based on the preset video display format, and restore the signal to a byte stream.

[0024] The scheduling instruction reassembler is used to reassemble the byte stream into the original scheduling instructions based on the preset video display format;

[0025] The instruction execution module executes the scheduling instructions to control the playback of each terminal that is unidirectionally connected to it.

[0026] According to some embodiments of the present invention, the frame encapsulation structure of the scheduling instruction includes: a frame header, a frame body, a line tail, and a frame tail; the frame body includes a line body, which serves as the actual carrier of the scheduling instruction; the pixel bearer mapper maps the scheduling instruction according to the pixel channel order, and the decoding and retiming module reads the pixel channel values ​​line by line, directly forming a byte stream according to the RGB corresponding 3 bytes, and restores the pixel channel values ​​to the byte stream.

[0027] According to some embodiments of the present invention, the scheduling instruction packer is further used to concatenate and encapsulate multiple scheduling instructions into an image frame. The frame header of the image frame includes an index table, which records the identity sequence number of each scheduling instruction, as well as the starting offset and length of the scheduling instruction in the image frame. The frame body structure arranges the scheduling instructions sequentially. After encapsulating an image frame, it is sent to the pixel bearer mapper or the encapsulated image frame is sent at predetermined time intervals.

[0028] According to some embodiments of the present invention, the unidirectional transmission link includes: an electro-optical conversion unidirectional optical transmitter for receiving audio and video signals and converting them into optical signals; a unidirectional optical fiber for transmitting the optical signals unidirectionally to an opto-optical conversion unidirectional optical receiver; and an opto-optical conversion unidirectional optical receiver for receiving the optical signals and converting them into electrical signals before outputting them.

[0029] According to some embodiments of the present invention, the terminal information includes terminal feature data, historical delivery effect data, and geographical location and scene data; the scheduling instructions include: playback time, playback content, push priority, Top-N list of best playback terminals, best coverage range, best coverage terminal for emergency content, instruction execution order, and multi-terminal synchronization strategy.

[0030] This invention enables automatic decision-making based on real-time terminal status, achieving optimal coupling between "content" and "terminal," significantly improving the targeting and reach of information dissemination, and solving the problems of inaccurate audience matching and delayed response in traditional manual scheduling methods. Furthermore, by combining unidirectional transmission links with preset pixel mapping formats, it reduces network communication protocol encapsulation overhead, prevents illegal protocol injection, ensures the security level of the intranet, and guarantees the real-time and stable delivery of instructions in high-time-demand scenarios such as emergency broadcasts, achieving automated control under physical security isolation within the intranet. Attached Figure Description

[0031] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used are briefly described below:

[0032] Figure 1 A flowchart illustrating an AI-based intelligent recommendation and automatic scheduling method for information transmission resources according to some embodiments of the present invention is shown.

[0033] Figure 2 A timing diagram is shown for an AI-based intelligent recommendation and automatic scheduling method for information transmission resources according to some embodiments of the present invention;

[0034] Figure 3 A block diagram of an AI-based intelligent recommendation and automatic scheduling system for information transmission resources according to some embodiments of the present invention is shown.

[0035] Figure 4 The diagram illustrates a data input unit and an AI intelligent analysis and scheduling unit in an AI-based intelligent recommendation and automatic scheduling system for information transmission resources, according to some embodiments of the present invention.

[0036] The correspondence between the reference numerals in the figure is as follows:

[0037] 100: Data Input Unit; 200: AI Intelligent Analysis and Scheduling Unit; 300: Scheduling Execution Unit; 310: Scheduling Instruction Packer; 3101: Pixel Bearer Mapper; 3102: Video Transmitter; 320: Unidirectional Transmission Link; 330: Intranet Terminal Control and Execution Module; 3301: Decoding and Retiring Module; 3302: Scheduling Instruction Reassembler; 3303: Instruction Execution Module; 210: Audience Profile Modeler; 220: Content Attribute Semantic Analyzer; 230: Geographic / Scene Hotspot Recognizer; 240: Intelligent Matching Engine; 250: Automatic Scheduling Strategy Generator. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0039] According to some embodiments of the present invention, an AI-based intelligent recommendation and automatic scheduling method for information transmission resources, such as... Figure 1 As shown, it includes the following steps:

[0040] S1. Input the terminal information and content attribute data;

[0041] The terminal information may include terminal feature data, historical delivery performance data, and geographic location and scene data. This data constitutes the foundational dataset for subsequent AI analysis.

[0042] Terminal characteristic data includes data such as audience, traffic flow, audience appearance time and scene; historical campaign performance data includes clicks, dwell time, viewing time, number of QR code scans, interaction heat distribution, audience sentiment recognition and conversion rate data.

[0043] Content attribute data includes the current content to be published (or the current information transmission resources to be published), such as news, emergency broadcasts, promotions, government publicity, commercial advertisements, building signage, weather and traffic information, interactive pages, etc.

[0044] Geographic location and scene data include data on public service areas such as buildings, shopping malls, subways, airports, high-speed rail, hospitals, and schools, as well as outdoor spaces such as scenic parks.

[0045] By incorporating multi-dimensional data such as audience information, historical effects, and geographical location, dispatch instructions are no longer broadcast blindly, but rather achieve precise delivery that is "tailored to local conditions and individual needs." For example, in scenarios such as military parades, inspections, or sudden emergencies, the system can automatically ensure that high-priority content is covered on the best terminals, thereby enhancing public safety capabilities.

[0046] S2. Analyze the information and content attribute data of each terminal through AI, and generate scheduling instructions based on the analysis results.

[0047] According to some embodiments of the present invention, the step of analyzing terminal information and content attribute data through AI and generating scheduling instructions based on the analysis results includes: generating audience profile vectors based on terminal feature data and historical delivery effect data, for example, generating audience profile vectors based on the location of the terminal (office, shopping mall, government service hall, etc.), traffic statistics at different time periods, audience changes during holidays / events, and historical delivery effect data; classifying content and extracting tags based on content attribute data to generate content semantic vectors; identifying hotspot areas based on geographical location and scene data; calculating recommendation scores based on audience profile vectors, content semantic vectors, and geographical / scene hotspot areas; and automatically generating the scheduling instructions based on the recommendation scores.

[0048] According to some embodiments of the present invention, AI can first generate audience profile vectors based on terminal feature data and historical delivery effect data. Simultaneously, it utilizes NLP or visual semantic models to model the attributes of the resource content to be published, extracting tags such as urgency, type (emergency broadcast / information / education / event promotion), effective time window, dissemination radius, or target audience to generate content semantic vectors. Furthermore, it can combine Geo-AI models to identify geographical / scene hotspots, including geographical hotspot identification: high foot traffic in commercial districts on weekends, peak hours in office buildings, and concentrated foot traffic in government service halls before and after Mondays / holidays; and time-based scene identification: class / leave, morning / evening peak hours, and holidays / event periods.

[0049] The scheduling instructions may include playback time, playback content, push priority, Top-N list of best playback terminals, best coverage range (coverage radius), best coverage terminals for emergency content (priority list), instruction execution order, and multi-terminal synchronization strategy.

[0050] By replacing the traditional manual scheduling with the above methods, the relevance of intranet content to the audience has been greatly improved, and the efficiency of resource allocation has been optimized.

[0051] S3. Based on a fixed preset video display format, the scheduling instructions are directly mapped to pixel channel values.

[0052] To achieve secure transmission across physical barriers, this invention does not employ traditional network protocols, but instead disguises scheduling commands as video signals. Simultaneously, a fixed output parameter (such as RGB 4:4:4, 8-bit, FullRange) is selected to ensure that pixel values ​​are not dynamically modified by system drivers or graphics APIs. The pixel-carrying mapper directly fills the binary byte stream of the scheduling commands into the R, G, and B channels of the pixels, achieving transparent data-to-pixel mapping.

[0053] According to some embodiments of the present invention, before the above direct mapping, the instruction can be encapsulated into a frame first. The frame encapsulation structure of the scheduling instruction includes: frame header, frame body, line tail, and frame tail; the frame body contains the line body, which serves as the actual carrier of the scheduling instruction; the scheduling instruction is mapped in pixel channel order, and the intranet terminal control and execution module reads the pixel channel values ​​line by line, directly forms a byte stream according to RGB corresponding to 3 bytes, and restores the pixel channel values ​​to the byte stream.

[0054] By defining "frame header, frame body, line end, and frame tail," a strict timing logic is provided for the originally disordered scheduling data, ensuring clear data boundaries in high-speed video streams. Furthermore, by utilizing the direct correspondence between RGB channels and 3 bytes, hardware-level transparent transmission is achieved, reducing the computational overhead of the intranet decoding module.

[0055] According to some embodiments of the present invention, multiple scheduling instructions can be concatenated and encapsulated into a single image frame. The frame header of the image frame includes an index table, which records the identity sequence number of each scheduling instruction, as well as the starting offset and length of the scheduling instruction within the image frame. The frame body structure arranges the scheduling instructions sequentially. After encapsulating an image frame, it is sent to a pixel bearer mapper or sent at predetermined time intervals. This allows multiple instructions to be carried within a single video frame via "concatenation" and managed using an index table, significantly improving data throughput per unit time and greatly enhancing transmission efficiency. Furthermore, the index table records the offset and length, allowing the receiving end to quickly locate and retrieve the target instruction based on the identity sequence number. Even if a single instruction has a partial error, it will not affect the reading of other instructions in the entire frame, enhancing fault tolerance and flexibility.

[0056] S4. Convert the pixel channel value into a pixel pass-through bearer signal and output it to a unidirectional transmission link;

[0057] The video transmitter converts the mapped pixel values ​​into pass-through signals that conform to display interface standards (such as HDMI and DP). During this process, all dynamic features that could alter pixel values ​​(such as color space conversion and sharpening) are disabled.

[0058] S5. The unidirectional transmission link transmits the pixel direct-pass bearer signal unidirectionally to the intranet terminal control and execution module.

[0059] According to some embodiments of the present invention, the unidirectional transmission link includes: an electro-optical conversion unidirectional optical transmitter for receiving audio and video signals and converting them into optical signals; a unidirectional optical fiber for transmitting the optical signals unidirectionally to an opto-optical conversion unidirectional optical receiver; and an opto-optical conversion unidirectional optical receiver for receiving the optical signals and converting them into electrical signals before outputting them.

[0060] The unidirectional transmission link employs photoelectric conversion technology at the physical layer. After electro-optical conversion, the signal is transmitted through a unidirectional optical fiber. The receiving end passively receives the signal without sending any handshake signals (such as ACK), thus completely severing the path of data transmission from the internal network to the external network at the physical layer, ensuring internal network security. The unidirectional optical fiber may include an optical shutter to achieve unidirectional optical transmission.

[0061] S6. The intranet terminal control and execution module decodes the received signal based on the preset video display format, re-times it, and restores the signal to a byte stream.

[0062] The decoding and retiming module on the intranet performs horizontal and vertical synchronization positioning on the electrical signal after photoelectric conversion according to a preset fixed video display format (such as a specific resolution and refresh rate), and restores the signal to a byte stream.

[0063] S7. Based on the preset video display format, reassemble the byte stream into the original scheduling instructions.

[0064] The scheduling instruction reassembler reads the pixel channel values ​​line by line, restoring the RGB values ​​to the original binary sequence every 3 bytes.

[0065] S8. The intranet terminal control and execution module executes the scheduling instructions to control the playback of each terminal that is unidirectionally connected to it according to the scheduling instructions.

[0066] For example, content can be precisely delivered to the corresponding terminals on the intranet based on content priority and a Top-N list of terminals.

[0067] Figure 2 The dynamic working sequence of the AI-based intelligent recommendation and automatic scheduling method for information transmission resources according to some embodiments of the present invention is shown.

[0068] The data input unit takes in information such as audience, content, geographic and historical data.

[0069] The AI ​​intelligent analysis and scheduling unit generates audience profile vectors; performs content semantic analysis to generate content semantic vectors; identifies geographical / scene hotspots; and automatically generates scheduling strategies based on the previous results, including the set of playback terminals, time, and priority; and then outputs automated scheduling instructions containing the scheduling strategy.

[0070] After obtaining the scheduling instruction, the scheduling instruction packager packages it and pushes it to the intranet via a one-way transmission link. This packaging process may include the aforementioned steps S3 and S4.

[0071] The intranet terminal control and execution module receives signals and converts them into scheduling instructions, automatically controlling the playback of each terminal to achieve real-time scheduling under secure isolation conditions.

[0072] This invention also provides an AI-based intelligent recommendation and automatic scheduling system for information transmission resources. For example... Figure 3 As shown, the system includes: a data input unit (100), an AI intelligent analysis and scheduling unit (200), and a scheduling execution unit (300).

[0073] The data input unit (100) is used to input terminal information and content attribute data.

[0074] The AI ​​intelligent analysis and scheduling unit (200) is connected to the data input unit (100). It analyzes the information and content attribute data of each terminal through AI and generates scheduling instructions based on the analysis results.

[0075] The scheduling execution unit (300) is connected to the AI ​​intelligent analysis and scheduling unit (200) and is used to execute the scheduling instructions.

[0076] The scheduling execution unit (300) includes a scheduling instruction packer (310), a one-way transmission link (320), and an intranet terminal control and execution module (330).

[0077] The scheduling instruction packer (310) includes a pixel bearer mapper (3101) and a video transmitter (3102); the intranet terminal control and execution module (330) includes a decoding and retiming module (3301), a scheduling instruction reassembler (3302), and an instruction execution module (3303).

[0078] The pixel carrier mapper (3101) is used to receive the scheduling instruction and directly map the scheduling instruction to pixel channel values ​​based on a fixed preset video display format.

[0079] The video transmitter (3102) is connected to the pixel bearer mapper (3101) for converting the pixel channel values ​​into pixel pass-through bearer signals and outputting them to the unidirectional transmission link (320).

[0080] The unidirectional transmission link (320) is used to transmit the pixel pass-through bearer signal unidirectionally to the decoding and retiming module (3301).

[0081] The decoding and re-timing module (3301) is used to decode and re-timing the received signal based on the preset video display format and restore the signal to a byte stream.

[0082] The scheduling instruction reassembler (3302) is used to reassemble the byte stream into the original scheduling instructions based on the preset video display format.

[0083] The instruction execution module (3303) executes the scheduling instruction to control the playback of each terminal that is unidirectionally connected to the intranet terminal control and execution module (330) according to the scheduling instruction.

[0084] According to some embodiments of the present invention, the terminal information acquired by the data input unit (100) includes: terminal feature data, historical delivery effect data, content attribute data, and geographic location and scene data. For details of the aforementioned data, please refer to the preceding descriptions surrounding the various figures.

[0085] According to some embodiments of the present invention, such as Figure 4 As shown, the AI ​​intelligent analysis and scheduling unit (200) includes: an audience profile modeler (210), which generates an audience profile vector based on terminal feature data and historical delivery effect data; a content attribute semantic analyzer (220), which classifies content and extracts tags based on content attribute data to generate a content semantic vector; a geographic / scene hotspot identifyer (230), which identifies hotspot areas based on geographic location and scene data; an intelligent matching engine (240), which fuses and matches the audience profile vector, content semantic vector and geographic / scene hotspot areas to calculate the recommendation score; and an automatic scheduling strategy generator (250), which automatically generates scheduling instructions based on the recommendation score.

[0086] According to some embodiments of the present invention, the intelligent matching engine (240) adopts a multi-source data fusion algorithm to fuse and match audience profile vectors, content semantic vectors and geographic / scene hotspot areas.

[0087] According to some embodiments of the present invention, the unidirectional transmission link (320) is, for example, a physically isolated link based on HDMI unidirectional optical transmission. According to some embodiments of the present invention, the unidirectional transmission link includes: an electro-optical conversion unidirectional optical transmitter for receiving audio and video signals and converting them into optical signals; a unidirectional optical fiber for unidirectionally transmitting the optical signals to an opto-optical conversion unidirectional optical receiver; and an opto-optical conversion unidirectional optical receiver for receiving the optical signals and converting them into electrical signals before outputting them.

[0088] According to some embodiments of the present invention, the frame encapsulation structure of the scheduling instruction includes: a frame header, a frame body, a line tail, and a frame tail; the frame body includes a line body, which serves as the actual carrier of the scheduling instruction; the pixel bearer mapper maps the scheduling instruction according to the pixel channel order, and the decoding and retiming module reads the pixel channel values ​​line by line, directly forming a byte stream according to the RGB corresponding 3 bytes, and restores the pixel channel values ​​to the byte stream.

[0089] According to some embodiments of the present invention, the scheduling instruction packer is further used to concatenate and encapsulate multiple scheduling instructions into an image frame. The frame header of the image frame includes an index table, which records the identity sequence number of each scheduling instruction, as well as the starting offset and length of the scheduling instruction in the image frame. The frame body structure arranges the scheduling instructions sequentially. After encapsulating an image frame, it is sent to the pixel bearer mapper or the encapsulated image frame is sent at predetermined time intervals.

[0090] In addition, the front is surrounded Figure 1 and 2 The same applies here, and will not be repeated for the sake of simplicity.

[0091] This invention enables automatic decision-making based on real-time terminal status, achieving optimal coupling between "content" and "terminal," significantly improving the targeting and reach of information dissemination, and solving the problems of inaccurate audience matching and delayed response in traditional manual scheduling methods. Furthermore, by combining unidirectional transmission links with preset pixel mapping formats, it reduces network communication protocol encapsulation overhead, prevents illegal protocol injection, ensures the security level of the intranet, and guarantees the real-time and stable delivery of instructions in high-time-demand scenarios such as emergency broadcasts, achieving automated control under physical security isolation within the intranet.

Claims

1. An AI-based intelligent recommendation and automatic scheduling method for information transmission resources, comprising the following steps: Input the terminal information and content attribute data; AI is used to analyze information and content attribute data from various terminals, and scheduling instructions are generated based on the analysis results. The scheduling instructions are directly mapped to pixel channel values ​​based on a fixed preset video display format, which includes: resolution, pixel format, and frame format. The pixel channel value is converted into a pixel pass-through bearer signal and output to a unidirectional transmission link; wherein, the pixel pass-through bearer signal is a pass-through signal conforming to the display interface standard, and all dynamic characteristics that may change the pixel value are disabled; The unidirectional transmission link transmits the pixel pass-through bearer signal unidirectionally to the intranet terminal control and execution module; The intranet terminal control and execution module decodes the received signal based on the preset video display format, re-times it, and restores the signal to a byte stream; it reassembles the byte stream into the original scheduling instructions based on the preset video display format; and it executes the scheduling instructions to control the playback of each terminal that is unidirectionally connected to it according to the scheduling instructions.

2. The method according to claim 1, characterized in that, The terminal information includes terminal feature data, historical delivery effect data, and geographical location and scene data; the scheduling instructions include: playback time, playback content, push priority, Top-N list of best playback terminals, best coverage range, best coverage terminals for emergency content, instruction execution order, and multi-terminal synchronization strategy.

3. The method according to claim 2, characterized in that, The method of using AI to analyze terminal information and content attribute data, and generating scheduling instructions based on the analysis results, includes: generating audience profile vectors based on terminal feature data and historical delivery effect data; generating content semantic vectors based on content attribute data; identifying geographical / scene hotspots; calculating recommendation scores based on audience profile vectors, content semantic vectors, and geographical / scene hotspots; and automatically generating the scheduling instructions based on the recommendation scores.

4. The method according to claim 1, characterized in that, The frame encapsulation structure of the scheduling instruction includes: frame header, frame body, line tail, and frame tail; the frame body contains the line body, which serves as the actual carrier of the scheduling instruction; the scheduling instruction is mapped in pixel channel order, and the intranet terminal control and execution module reads the pixel channel values ​​line by line, directly forming a byte stream according to RGB corresponding to 3 bytes, and restoring the pixel channel values ​​to the byte stream.

5. The method according to claim 1, characterized in that, Multiple scheduling instructions are concatenated and encapsulated into an image frame. The header of the image frame contains an index table, which records the identity sequence number of each scheduling instruction, as well as the starting offset and length of the scheduling instruction in the image frame. The frame body structure arranges the scheduling instructions in sequence. After an image frame is encapsulated, it is sent to the pixel bearer mapper or sent at predetermined time intervals.

6. An AI-based intelligent recommendation and automatic scheduling system for information transmission resources, characterized in that: include: Data input unit, AI intelligent analysis and scheduling unit, and scheduling execution unit; The data input unit is used to input terminal information and content attribute data; The AI ​​intelligent analysis and scheduling unit is connected to the data input unit. It analyzes the information and content attribute data of each terminal through AI and generates scheduling instructions based on the analysis results. The scheduling execution unit is connected to the AI ​​intelligent analysis and scheduling unit and is used to execute the scheduling instructions; The scheduling execution unit includes a scheduling instruction packager, a one-way transmission link, and an intranet terminal control and execution module; The scheduling instruction packer includes a pixel bearer mapper and a video transmitter; The intranet terminal control and execution module includes a decoding and retiming module, a scheduling instruction reassembler, and an instruction execution module; The pixel bearer mapper is used to receive the scheduling instruction and directly map the scheduling instruction into pixel channel values ​​based on a fixed preset video display format; Preset video display formats include: resolution, pixel format, and frame format; The video transmitter is connected to the pixel bearer mapper, which is used to convert the pixel channel value into a pixel pass-through bearer signal and output it to a one-way transmission link. The pixel pass-through bearer signal is a pass-through signal that conforms to the display interface standard, and all dynamic features that may change the pixel value are disabled. The unidirectional transmission link is used to transmit the pixel pass-through bearer signal unidirectionally to the decoding and retiming module; The decoding and retiming module is used to decode the received signal, retime it based on the preset video display format, and restore the signal to a byte stream. The scheduling instruction reassembler is used to reassemble the byte stream into the original scheduling instructions based on the preset video display format; The instruction execution module executes the scheduling instructions to control the playback of each terminal that is unidirectionally connected to it.

7. The system according to claim 6, characterized in that, The frame encapsulation structure of the scheduling instruction includes: frame header, frame body, line tail, and frame tail; the frame body contains the line body, which serves as the actual carrier of the scheduling instruction; the pixel carrier mapper maps the scheduling instruction according to the pixel channel order, and the decoding and retiming module reads the pixel channel values ​​line by line, directly forming a byte stream according to the RGB corresponding 3 bytes, and restores the pixel channel values ​​to the byte stream.

8. The system according to claim 7, characterized in that, The scheduling instruction packer is also used to concatenate and encapsulate multiple scheduling instructions into an image frame. The frame header of the image frame contains an index table, which records the identity sequence number of each scheduling instruction, as well as the starting offset and length of the scheduling instruction in the image frame. The frame body structure arranges each scheduling instruction in sequence. After encapsulating an image frame, it is sent to the pixel bearer mapper or the encapsulated image frame is sent at a predetermined time interval.

9. The system according to claim 8, characterized in that, The unidirectional transmission link includes: an electro-optical conversion unidirectional optical transmitter, which receives audio and video signals and converts them into optical signals; a unidirectional optical fiber, which transmits the optical signals unidirectionally to an opto-optical conversion unidirectional optical receiver; and an opto-optical conversion unidirectional optical receiver, which receives the optical signals and converts them into electrical signals before outputting them.

10. The system according to claim 6, characterized in that, The terminal information includes terminal feature data, historical delivery effect data, and geographical location and scene data; the scheduling instructions include: playback time, playback content, push priority, Top-N list of best playback terminals, best coverage range, best coverage terminals for emergency content, instruction execution order, and multi-terminal synchronization strategy.

Images