Media control system of exhibition hall

Abstract

The invention relates to the technical field of exhibition hall media control, discloses an exhibition hall media control system, and aims to solve the problems of single interaction, resource waste and high energy consumption of an existing system. The system comprises a sensing layer, a transmission layer, a processing layer, an application layer and a collaboration mechanism. Three-dimensional user portraits of identities, behaviors and interests are constructed based on multi-dimensional perception, personalized content pushing is achieved in combination with a collaborative filtering algorithm, cross-exhibition-area content connection and lamplight and sound equipment collaboration are triggered through UWB high-precision positioning, coherent immersive experience is provided for users, and the user experience is improved. Meanwhile, server computing power is dynamically distributed by utilizing an LSTM (Long Short Term Memory) people flow prediction model and edge computing, a core server and edge node computing power ratio is optimized at peak time, dormancy redundancy computing power is realized at low peak time, and multi-condition linkage low-carbon energy-saving logic can be constructed based on people flow density, natural illumination and equipment non-interaction duration, so that the comprehensive energy consumption of the system is reduced, and the energy-saving performance of the system is improved. And user experience interruption caused by energy saving is avoided through a quick wake-up mechanism.

Description

Technical Field

[0001] This invention relates to the field of exhibition hall media control technology, and more particularly to an exhibition hall media control system. Background Technology

[0002] In museums, science and technology museums, and corporate brand showrooms, media control systems serve as the core carriers for content display and user interaction. Their performance directly impacts user experience, resource utilization efficiency, and operating costs. While existing showroom media control systems have achieved basic video playback and lighting control functions with the development of digital technology, they still suffer from three key technical deficiencies in practical applications, making it difficult to meet the current demands of showrooms for personalized experiences, efficient resource utilization, and low-carbon operation. These deficiencies are as follows: Firstly, existing systems suffer from a lack of interaction and content coherence, failing to provide a personalized and immersive experience. Current exhibition hall media control systems mostly employ a passive mode of fixed content looping and manual touchscreen triggering. This lacks both an effective user data collection and analysis mechanism and cross-exhibition area content linkage capabilities. On one hand, the system can only provide feedback to users through simple buttons or touchscreens, failing to generate personalized content recommendations based on user identity, behavioral characteristics, and interests. This results in different users receiving the same information, leading to a fragmented user experience. On the other hand, when users move between different exhibition areas, the media content, lighting, and sound systems in each area operate independently, hindering content continuity and equipment coordination. This causes a break in the exhibition hall's narrative chain, making it difficult to create an immersive experience.

[0003] Secondly, the existing system's resource scheduling mechanism is rigid, resulting in peak configuration redundancy and off-peak resource waste. The flow of people in the exhibition hall exhibits significant tidal fluctuations, but the existing media control system's server computing power allocation and content loading strategies are based on fixed configurations for the maximum peak flow. On the one hand, the core server needs to maintain full computing power for a long time to cope with peak demand, while during off-peak periods, the server still runs at 100% computing power, resulting in a computing power idle rate of over 50%, causing a waste of energy and hardware resources. On the other hand, media content is stored on the core server, and users need to retrieve it from the core server in real time when accessing it. There is no edge preloading based on the content access frequency, which not only causes content loading to lag during peak periods due to excessive data transmission, but also further increases the burden on the core server, forming a double contradiction of resource configuration redundancy and poor access experience.

[0004] Third, the existing system's energy consumption control methods are crude and struggle to balance low-carbon operation with user experience. Current exhibition hall media control systems often employ a one-size-fits-all approach to energy management. Either they prioritize display effects by running screens, lights, and sound equipment at 100% power for extended periods, resulting in high energy consumption; or they resort to simple power-off strategies to reduce energy consumption, but this leads to a lack of content and insufficient ambient light when users enter the exhibition area, thus damaging the user experience. Existing systems lack an energy consumption optimization logic that considers multiple factors such as visitor density, natural lighting, and inactivity time, failing to find a balance between energy saving and user experience. This not only fails to meet current low-carbon operation requirements but also makes it difficult to adapt to the dynamically changing usage scenarios of exhibition halls.

[0005] Therefore, there is an urgent need for a new type of media control system with multi-dimensional perception, dynamic resource scheduling and cross-domain collaboration capabilities to solve the above-mentioned technical problems. Summary of the Invention

[0006] One objective of this invention is to propose a media control system for exhibition halls. This invention can construct a three-dimensional user profile based on multi-dimensional perception of identity, behavior, and interests, and combine it with a collaborative filtering algorithm to achieve personalized content push. It also uses UWB high-precision positioning to trigger cross-exhibition area content connection and the coordination of lighting and sound equipment to provide users with a seamless immersive experience. At the same time, it utilizes an LSTM crowd flow prediction model and edge computing to dynamically allocate server computing power, optimize the computing power ratio of core servers and edge nodes during peak hours, and hibernate redundant computing power during off-peak hours. Combined with a high-frequency content edge preloading mechanism, it can improve resource utilization efficiency. Furthermore, it can construct a low-carbon and energy-saving logic based on multiple conditions such as crowd density, natural lighting, and device inactivity time. While reducing the overall energy consumption of the system, it avoids user experience interruption caused by energy saving through a rapid wake-up mechanism, effectively breaking through the technical bottlenecks of existing exhibition hall media control systems.

[0007] A media control system for an exhibition hall according to an embodiment of the present invention includes a perception layer, a transmission layer, a processing layer, an application layer, and a coordination mechanism; The perception layer includes a multi-dimensional perception subsystem for collecting user data, environmental data, and device status data. The transmission layer includes a 5G+ edge computing transmission subsystem, which is used to achieve low-latency, high-bandwidth data transmission and edge preprocessing; The processing layer includes an intelligent control core subsystem for performing AI decision-making, dynamic resource scheduling, and cross-exhibition area linkage logic operations. The application layer includes a media output and interaction subsystem, which is used to display media content and enable user interaction; The collaborative mechanism includes a cross-exhibition area linkage protocol and a low-carbon energy-saving engine, which are used to realize collaborative control between exhibition areas and dynamic optimization of equipment energy consumption, respectively.

[0008] Furthermore, the multi-dimensional perception subsystem includes a user data acquisition module, which comprises an identity recognition unit, a behavior perception unit, and an interest analysis unit. The identity recognition unit is equipped with a 2-megapixel binocular camera and an RFID wristband that supports non-contact recognition within 10 meters, used to obtain user identity information; The behavior perception unit is equipped with an infrared human body sensor with a detection distance of 0.1-10m, a millimeter-wave radar with an identification accuracy of ±0.5m, and a touch screen interactive log collector to record user behavior data. The interest analysis unit is equipped with an NLP algorithm to parse user voice interaction content and extract interest keywords.

[0009] Furthermore, the multi-dimensional perception subsystem also includes an environmental data acquisition module and a device status acquisition module; The environmental data acquisition module is equipped with a temperature and humidity sensor with a measurement range of 0-60℃ and an accuracy of ±0.5℃, a light sensor with a measurement range of 0-100000 lux and an accuracy of ±5%, and a crowd density sensor with a counting error of ≤3%. The device status acquisition module is used to collect the power consumption, operating temperature and fault codes of the media device in real time, and upload them to the processing layer via the LoRa protocol with a transmission distance of ≤1km.

[0010] Furthermore, the 5G+ edge computing transmission subsystem includes a transmission network and edge nodes; The transmission network adopts a 5G private network with a downlink rate of ≥1Gbps, an uplink rate of ≥100Mbps and a latency of ≤10ms, and is configured with Ethernet backup; Each edge node is one edge server deployed in each exhibition area; The edge nodes can perform data preprocessing, local emergency control in the event of core network failure, and direct issuance of commands with high real-time requirements.

[0011] Furthermore, the intelligent control core subsystem includes a user profile engine, which constructs a three-dimensional user profile based on data collected from the perception layer, including identity, behavior, and interests. It then uses a collaborative filtering algorithm that combines user similarity and content similarity to generate a personalized content recommendation list, and the update frequency of the recommendation list is ≤10 seconds.

[0012] Furthermore, the intelligent control core subsystem also includes a dynamic resource scheduling module, which includes a server resource scheduling unit and a media resource scheduling unit; The server resource scheduling unit uses an LSTM model to predict the crowd density in each exhibition area over the next 30 minutes and dynamically adjusts the computing power allocation between the core server and edge nodes. During peak hours, the core server handles 70% of the computing power and the edge nodes handle 30%. During off-peak hours, the core server sleeps with 50% of the computing power and the edge nodes handle 50%. The media resource scheduling unit can establish a content cache pool that supports 4K / 8K video and 3D models. When the content is accessed ≥5 times in the past hour, the content is preloaded to the edge node.

[0013] Furthermore, the intelligent control core subsystem also includes a cross-exhibition area linkage control module. This module tracks the user's location based on UWB positioning technology with a positioning accuracy of ±10cm. When the user moves from exhibition area A to exhibition area B, a content connection protocol is triggered, causing exhibition area B to play a sequel related to the content in exhibition area A. At the same time, a device collaboration protocol is triggered, causing the lights in exhibition area A to decrease from 100% to 30% within 5 seconds, and the lights in exhibition area B to increase from 30% to 100% within 5 seconds. Simultaneously, the audio signal is switched to the audio signal of exhibition area B.

[0014] Furthermore, the low-carbon energy-saving engine can turn off 50% of the screens and keep one core screen playing the summary in a loop when the density of people in the exhibition area is ≤0.1 people / ㎡, while reducing the brightness of the lights to 50%; the low-carbon energy-saving engine can reduce the screen brightness to 70% and turn off the artificial lighting on the same side when the natural light is ≥5000 lux; the low-carbon energy-saving engine can enter a low-power mode with a power consumption of 20% of the normal mode when the device has no interaction time for ≥15 minutes, and wake up the device within 1 second when a user approaches to a distance of ≤2m.

[0015] Furthermore, the content output module of the media output and interaction subsystem supports multi-terminal adaptation and multiple content formats; Multi-terminal compatibility includes LED splicing screens that support up to 100 screens, transparent OLED screens with a transparency of ≥40%, and AR glasses that support spatial anchoring and can overlay 3D virtual exhibits. Multiple content formats are available, including 4K / 8K video, 3D models that support gesture control for rotation / zoom, and VR panoramas with a 360° field of view.

[0016] Furthermore, the user interaction module of the media output and interaction subsystem includes a multimodal interaction unit and a personalized customization unit; The multimodal interaction unit supports Chinese and English voice interaction, gesture interaction that recognizes six gestures (click, swipe, zoom, rotate, fist, and wave) through a TOF camera, and AR gesture grasping for virtual grabbing of exhibit models. The personalized customization unit allows users to save content via touchscreen. The system generates a unique QR code, which users can scan to obtain a content link valid for 72 hours.

[0017] The beneficial effects of this invention are: 1. This invention constructs a comprehensive user profile encompassing identity, behavior, and interests through the collaboration of a multi-dimensional perception subsystem and an intelligent control core. It utilizes a 2-megapixel binocular camera and RFID wristband to confirm user identity, millimeter-wave radar to track behavioral trajectories, and NLP algorithms to extract interest keywords from voice interactions. Furthermore, a collaborative filtering algorithm that integrates user and content similarity pushes personalized content to different users. UWB positioning technology triggers cross-exhibition area linkage, automatically connecting related content as users move, and simultaneously achieving coordinated switching of lighting and sound. This completely breaks the limitations of existing systems where exhibition areas are isolated and content is fixed, significantly improving user interaction satisfaction and greatly enhancing the accuracy of personalized push notifications.

[0018] 2. In this invention, to address the resource idleness problem caused by the tidal fluctuations in visitor flow in exhibition halls, the system collaborates with a 5G+edge computing transmission subsystem and a dynamic resource scheduling module. Edge nodes first perform H.265 encoding preprocessing on the perception layer data to reduce the transmission pressure on the core server. The processing layer uses an LSTM model to predict visitor density 30 minutes in advance. During peak hours, the core server undertakes 70% of the computing power, and the edge nodes undertake 30%. During off-peak hours, the core server sleeps with 50% of the computing power. At the same time, a content cache pool is established, and content with ≥5 visits in the past hour is automatically preloaded to the edge nodes. This mechanism greatly improves the utilization rate of server resources and speeds up content loading, avoiding the problem of existing systems having fixed configurations based on peak hours and over 50% resource idleness during off-peak hours.

[0019] 3. In this invention, the system interacts in real time with the low-carbon energy-saving engine through a multi-dimensional perception subsystem to construct a triple linkage optimization logic. When the flow density is ≤0.1 people / ㎡, 50% of the auxiliary screen is turned off and the lighting is reduced to 50%. When the natural light is ≥5000 lux, the screen brightness is reduced to 70% and the lighting on the same side is turned off. When the device has no interaction for ≥15 minutes, it enters a low-peak mode with 20% power consumption. It is quickly woken up when the user approaches. This data-driven precise control not only avoids the high energy consumption problem of the existing system running at full load, but also solves the defect of the traditional energy-saving solution that causes the experience to be interrupted due to power failure, thus achieving a balance between energy saving and experience. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the overall framework of a media control system for an exhibition hall proposed in this invention. Detailed Implementation

[0021] To make the technical means and objectives and effects of the present invention easier to understand, the embodiments of the present invention will be described in detail below with reference to specific illustrations.

[0022] Example 1 like Figure 1 As shown, this invention discloses a media control system for an exhibition hall, including a perception layer, a transmission layer, a processing layer, an application layer, and a collaboration mechanism; Specifically, the perception layer includes a multi-dimensional perception subsystem for collecting user data, environmental data, and device status data; The transport layer includes a 5G+ edge computing transport subsystem, which is used to achieve low-latency, high-bandwidth data transmission and edge preprocessing; The processing layer includes an intelligent control core subsystem for performing AI decision-making, dynamic resource scheduling, and cross-exhibition area linkage logic operations; The application layer includes a media output and interaction subsystem, which is used to display media content and enable user interaction; the collaboration mechanism includes a cross-exhibition area linkage protocol and a low-carbon energy-saving engine, which are used to achieve collaborative control between exhibition areas and dynamic optimization of equipment energy consumption, respectively.

[0023] In actual deployment, each sub-module of the perception layer connects to the transmission layer via the LoRa wireless protocol or gigabit Ethernet. The 5G private network of the transmission layer adopts an independent networking architecture and establishes a dedicated data channel with the core server of the processing layer. The processing layer sends control commands to the application layer via the TCP / IP protocol. The coordination mechanism realizes data interaction with each layer through the middleware of the processing layer. For example, the cross-exhibition area linkage protocol needs to obtain UWB positioning data from the processing layer in real time, and the low-carbon energy-saving engine needs to receive the crowd density and light data uploaded by the perception layer to trigger the energy consumption optimization logic.

[0024] The multi-dimensional perception subsystem includes a user data acquisition module, which comprises an identity recognition unit, a behavior perception unit, and an interest analysis unit. The identity recognition unit is equipped with a 2-megapixel binocular camera and an RFID wristband that supports non-contact identification within 10 meters, used to obtain user identity information; The behavior perception unit is equipped with an infrared human body sensor with a detection range of 0.1-10m, a millimeter-wave radar with an identification accuracy of ±0.5m, and a touch screen interactive log collector to record user behavior data. The interest analysis unit is equipped with an NLP algorithm to parse user voice interaction content and extract interest keywords.

[0025] In practice, the binocular camera of the identity recognition unit is installed 2.8m above the ground directly above the entrance of the exhibition hall. It can automatically capture the facial images of users entering the hall and compare them with the exhibition hall reservation database. If the match is successful, identity labels such as "child (6-12 years old)", "adult (18-60 years old)" and "expert (holding relevant field certificates)" are generated. If no facial information is matched, the system will guide the user to wear an RFID wristband through a prompt at the entrance. The unique ID built into the wristband will be associated with a temporary identity tag for "ordinary tourists".

[0026] The infrared human body sensor of the behavior perception unit is installed 1.5m above the ground in front of each exhibit, with a response time of ≤0.3 seconds, and can detect whether the user is standing in front of the exhibit; the millimeter-wave radar is installed in the center of the top of the exhibition area to record the user's movement trajectory and dwell time in real time. The touchscreen interactive log collector is integrated into the 15.6-inch capacitive touchscreen in each exhibition area, recording user clicks, playback, favorites and other operations, and uploading the logs to the edge node every 30 seconds.

[0027] The NLP algorithm of the interest analysis unit is based on the BERT-Chinese pre-trained model. It collects user voice through microphones in the exhibition hall, and the algorithm automatically parses the voice content, extracts interest keywords such as steam engine and working principle, and matches them with keywords in the content database to generate interest tags.

[0028] Meanwhile, the multi-dimensional perception subsystem also includes an environmental data acquisition module and an equipment status acquisition module; The environmental data acquisition module is equipped with a temperature and humidity sensor with a measurement range of 0-60℃ and an accuracy of ±0.5℃, a light sensor with a measurement range of 0-100000 lux and an accuracy of ±5%, and a crowd density sensor with a counting error of ≤3%. The device status acquisition module is used to collect the power consumption, operating temperature and fault codes of media devices in real time, and upload them to the processing layer via the LoRa protocol with a transmission distance of ≤1km.

[0029] During actual installation, one temperature and humidity sensor is installed for every 50 square meters of exhibition area, fixed on the side wall of the exhibition area 1.2m above the ground, and collects data every minute. If the detected temperature is ≥30℃ and the humidity is ≥60%, the system will link the air conditioning equipment to adjust the environmental parameters. The light sensor is installed at the top of the exhibition area 0.5m above the window, and collects the natural light intensity every minute. The data is directly used for the brightness adjustment logic of the low-carbon energy-saving engine. The crowd density sensor counts the number of people in the exhibition area based on the YOLOv8 target detection algorithm, with a counting error of ≤3%, and uploads the crowd data to the processing layer every 2 minutes.

[0030] The device status acquisition module collects the real-time power consumption of the media device through the smart power socket and the operating temperature through the temperature sensor on the device motherboard. If the device malfunctions, it will automatically generate a fault code. All data is uploaded to the processing layer through the LoRa module. After receiving the fault code, the processing layer will send an alarm message to the mobile phone of the maintenance personnel through the exhibition hall management backend.

[0031] The transport layer includes the 5G+ edge computing transport subsystem, which comprises the transport network and edge nodes; The transmission network adopts a 5G private network with a downlink rate of ≥1Gbps, an uplink rate of ≥100Mbps and a latency of ≤10ms, and is configured with Ethernet backup; the edge node is one edge server deployed in each exhibition area; the edge node can perform data preprocessing, local emergency control in case of core network failure, and direct issuance of commands with high real-time requirements.

[0032] In practice, the 5G private network adopts an independent networking architecture, with core network equipment deployed in the exhibition hall's computer room to ensure downlink speeds ≥1Gbps and uplink speeds ≥100Mbps. Through network testing, data transmission latency is ≤10ms. Ethernet backup uses link aggregation technology to bind two gigabit links. When the 5G private network fails, the system automatically switches to Ethernet transmission with a switching time of ≤1 second.

[0033] Each exhibition area is equipped with one edge server. The data preprocessing function of the edge nodes uses the FFmpeg tool library to implement H.265 encoding, compressing the 1080P video collected by the perception layer to 0.8Mbps before uploading it to the processing layer, reducing the amount of data transmission. In the event of a core network failure, the edge nodes will temporarily take over the media control of their respective exhibition areas, ensuring only basic functions and not involving cross-exhibition area linkage. For commands with high real-time requirements, the edge nodes directly send them to the device controller via the RS485 protocol, with a response latency of ≤50ms.

[0034] The processing layer includes a core intelligent control subsystem, which is a user profile engine. The user profile engine constructs a three-dimensional user profile based on the data collected by the perception layer, including identity, behavior, and interests. It uses a collaborative filtering algorithm that combines user similarity and content similarity to generate a personalized content recommendation list, and the recommendation list is updated every 10 seconds.

[0035] In actual operation, the user profiling engine is implemented using Python's Scikit-learn library. It first integrates identity tags (such as "children aged 6-12"), behavior tags (such as "stay time ≥ 5 minutes (deep browsing)" and "clicked video ≥ 3 times (video preference)"), and interest tags (such as "pre-Qin history" and "aerospace technology") uploaded by the perception layer to construct a three-dimensional user profile instance, such as "identity: children aged 6-12, behavior: deep browsing + video preference, interest: dinosaur fossils".

[0036] The implementation steps of the collaborative filtering algorithm are as follows: The first step is to calculate user similarity based on the Pearson correlation coefficient; The second step is to extract keywords from the media content based on the TF-IDF algorithm and calculate the content similarity. The third step is to combine user similarity and content similarity to generate a recommendation score, and select the top 3 scores to form a recommendation list.

[0037] The recommendation list is updated frequently using a timer. Every 10 seconds, the latest user behavior data from the perception layer is read, the recommendation list is recalculated, and then sent to the application layer, where it is displayed as a "Recommended for You" section at the top of the screen in front of the user.

[0038] In addition, the intelligent control core subsystem also includes a dynamic resource scheduling module, which includes a server resource scheduling unit and a media resource scheduling unit. The server resource scheduling unit uses an LSTM model to predict the density of people in each exhibition area in the next 30 minutes and dynamically adjusts the computing power allocation between the core server and the edge nodes. During peak hours, the core server undertakes 70% of the computing power and the edge nodes undertake 30%. During off-peak hours, the core server sleeps with 50% of the computing power and the edge nodes undertake 50%. The media resource scheduling unit can establish a content cache pool that supports 4K / 8K video and 3D models. When the content is accessed ≥5 times in the past hour, the content is preloaded to the edge nodes.

[0039] In practice, the LSTM model is trained using the TensorFlow framework. The training data consists of hourly pedestrian density data from the past 30 days. Input features include time, date type, and weather conditions. The output is a predicted pedestrian density for the next 30 minutes. The model is validated using data from the past 7 days as a test set, and the prediction accuracy is ≥85%.

[0040] Server resource scheduling is achieved through the Kubernetes container orchestration platform, and computing power allocation is based on CPU utilization: during peak traffic periods, the CPU utilization of core servers is controlled at 70%-80%, and the CPU utilization of edge nodes is controlled at 60%-70%; during off-peak periods, core servers shut down 50% of their cores through CPU hibernation technology, and the CPU utilization is controlled at 30%-40%, while the CPU utilization of edge nodes is controlled at 50%-60%.

[0041] The media resource scheduling unit establishes a content cache pool in the storage array of the core server, allocating a cache space of 500GB / area. The edge nodes count the number of content accesses every hour. When the number of accesses is ≥5, the processing layer sends a preloading instruction to the corresponding edge node. The edge node downloads the content from the core server through the 5G private network and stores it on the local SSD. The next time the user accesses the content, it can be loaded directly from the edge node, improving the loading speed by more than 50%.

[0042] The core intelligent control subsystem also includes a cross-exhibition area linkage control module. This module tracks the user's location based on UWB positioning technology with a positioning accuracy of ±10cm. When the user moves from exhibition area A to exhibition area B, a content connection protocol is triggered, causing exhibition area B to play a sequel related to the content in exhibition area A. At the same time, a device collaboration protocol is triggered, causing the lights in exhibition area A to decrease from 100% to 30% within 5 seconds, and the lights in exhibition area B to increase from 30% to 100% within 5 seconds. The audio signal is also switched to that of exhibition area B.

[0043] In actual deployment, the UWB positioning system uses Decawave DW1000 modules, with 4 UWB anchor points deployed in each exhibition area, and the RFID wristbands worn by users integrate UWB tags; The criteria for determining cross-exhibition zone movement are: if a UWB tag moves from the anchor point coverage area of ​​exhibition zone A into the anchor point coverage area of ​​exhibition zone B for 10 seconds without reverting, the system determines that the user has moved across exhibition zones. The content connection protocol is based on the exhibition hall content map. For example, if exhibition zone A is for pre-Qin pottery and plays a video on pre-Qin pottery making techniques, and exhibition zone B is for Qin and Han pottery and its associated follow-up content is a video on the evolution of pottery forms from the pre-Qin to the Qin and Han dynasties, when a user enters exhibition zone B, the processing layer sends a playback command to the screen of exhibition zone B with a delay of ≤1 second.

[0044] The device collaboration protocol controls the lighting dimmer via the DMX512 protocol. The lights in exhibition area A linearly decrease from 100% brightness to 30% in 5 seconds; the lights in exhibition area B linearly increase from 30% to 100% in 5 seconds. The audio switching is achieved through an audio matrix. At the same time as the lighting adjustment begins, the audio signal of exhibition area A is cut off and the local audio file of exhibition area B is connected. The switching is noise-free and the synchronization error is ≤200ms.

[0045] The collaborative mechanism includes a low-carbon energy-saving engine that can turn off 50% of the screens and keep one core screen playing the summary in a loop when the density of people in the exhibition area is ≤0.1 people / ㎡, while reducing the brightness of the lights to 50%; the low-carbon energy-saving engine can reduce the screen brightness to 70% and turn off the artificial lighting on the same side when the natural light is ≥5000 lux; the low-carbon energy-saving engine can enter a low-power mode with a power consumption of 20% of the normal mode when the device has no interaction time for ≥15 minutes, and wake up the device within 1 second when a user approaches to a distance of ≤2m.

[0046] In practice, the criteria for determining a crowd density of ≤0.1 people / ㎡ are as follows: the crowd density sensor counts the number of people in the exhibition area every 5 minutes, and optimization is triggered if the count is met twice consecutively; when 50% of the screens are turned off, auxiliary screens at the edges of the exhibition area are turned off first, while the core screen in the center of the exhibition area is kept on, and the core screen plays a summary of the exhibition area's theme in a loop. For example, in the Pre-Qin Culture exhibition area, which includes 10 key exhibits such as pottery and bronzes, the recommended visit time is 15 minutes, with a playback time of 3 minutes per cycle; the lighting brightness is reduced to 50% through a dimmer, with the brightness value reduced from 1000 lux to 500 lux. The detection of natural light ≥5000 lux is achieved through a light sensor, which detects the light every 1 minute. If the detection value is ≥5000 lux for 3 consecutive times, the screen brightness is reduced from 500 cd / ㎡ to 350 cd / ㎡, and the LED spotlights on the same side as the windows are turned off, while the spotlights on the other side are kept on to maintain basic lighting. The device's inactivity time is counted from the last user interaction. When the inactivity time is ≥15 minutes, the device enters a low-power mode: the screen brightness is reduced to 100 cd / ㎡, the speaker volume is reduced to 0, the server CPU frequency is reduced from 3.5GHz to 1.5GHz, and the power consumption is reduced from 300W in normal mode to 60W. When the infrared human body sensor detects that the user is close to within ≤2m, a wake-up command is immediately sent, and the device returns to normal mode within 1 second.

[0047] The content output module of the media output and interaction subsystem included in the application layer supports multi-terminal adaptation and multiple content formats; Multi-terminal compatibility includes LED splicing screens that support up to 100 screens, transparent OLED screens with a transparency of ≥40%, and AR glasses that support spatial anchoring and can overlay 3D virtual exhibits; multiple content formats include 4K / 8K videos, 3D models that support gesture control for rotation / zoom, and VR panoramas with a 360° field of view.

[0048] In practical applications, LED video walls can be spliced ​​up to 100 screens using a splicing controller to display 8K ultra-high-definition video.

[0049] The transparent OLED screen uses the LG55EG9A7V model, with a transparency of 45%. It is installed on the outside of the glass display case and can overlay 3D virtual models of the artifacts inside the display case. Users can see the superimposed effect of the real objects and virtual models through the screen.

[0050] AR glasses use SLAM technology to overlay 3D virtual exhibits onto the actual space of the exhibition hall, allowing users to observe the virtual exhibits from different angles after wearing the glasses.

[0051] Among the multiple content formats, 4K / 8K videos are stored in the video server of the core server and streamed to the terminal via the RTSP protocol; 3D models are in GLB format and developed using the Unity engine, allowing users to interact with the models through gesture control; VR panoramas are captured using an Insta360Pro 28K panoramic camera and processed into an interactive 360° panoramic view using Pano2VR software, allowing users to view different angles of the panorama by swiping on the touchscreen or using gestures.

[0052] The user interaction module of the media output and interaction subsystem includes a multimodal interaction unit and a personalized customization unit; The multimodal interaction unit supports Chinese and English voice interaction, gesture interaction that recognizes six gestures (click, swipe, zoom, rotate, fist, and wave) through a TOF camera, and AR gesture grasping for virtual grabbing of exhibit models; the personalized customization unit allows users to save content via touchscreen, and the system generates a unique QR code that users can scan to obtain a content link valid for 72 hours.

[0053] In practice, the voice interaction of the multimodal interaction unit adopts the intelligent cloud voice recognition API, which supports Chinese and English. The wake word is set to the exhibition hall assistant by default, and users can customize the wake word on the touch screen. Through noise testing, in an exhibition hall environment with noise ≤40dB, the voice recognition accuracy is ≥98%.

[0054] The TOF camera is installed above the screen at a height of 2m from the ground and can recognize 6 gestures: tap (index finger quickly taps the air once), swipe (index finger swipes horizontally to the left / right ≥10cm), zoom (thumb and index finger move from together to apart ≥5cm, or from apart to together), rotate (index finger rotates around thumb clockwise / counterclockwise ≥90°), clench fist (palm fully closed ≥1 second), and wave (palm waves left and right ≥20cm). The gesture recognition response time is ≤0.5 seconds.

[0055] AR gesture grasping is achieved through the HoloLens 2's hand tracking function. After the user wears the glasses, the system recognizes the user's hand movements, such as opening the palm and pointing it at the virtual model, generating a virtual grasping effect. The user can drag the virtual model to their eyes to view the details.

[0056] In the personalized customization unit, after the user clicks the favorite button on the touch screen, the system generates a QR code using Python's qrcode library and displays the QR code on the screen; After scanning the QR code with their mobile phones, users are redirected to a dedicated WeChat mini-program in the exhibition hall. The mini-program stores a link to the content, and the link is valid from the time the QR code is generated. It will automatically expire after 72 hours. Users can repeatedly view the saved content within the validity period.

[0057] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A media control system for an exhibition hall, characterized in that, It includes the perception layer, transmission layer, processing layer, application layer, and coordination mechanism; The perception layer includes a multi-dimensional perception subsystem for collecting user data, environmental data, and device status data. The transmission layer includes a 5G+ edge computing transmission subsystem, which is used to achieve low-latency, high-bandwidth data transmission and edge preprocessing; The processing layer includes an intelligent control core subsystem for performing AI decision-making, dynamic resource scheduling, and cross-exhibition area linkage logic operations. The application layer includes a media output and interaction subsystem, which is used to display media content and enable user interaction; The collaborative mechanism includes a cross-exhibition area linkage protocol and a low-carbon energy-saving engine, which are used to realize collaborative control between exhibition areas and dynamic optimization of equipment energy consumption, respectively.

2. The media control system for an exhibition hall according to claim 1, characterized in that, The multi-dimensional perception subsystem includes a user data acquisition module, which comprises an identity recognition unit, a behavior perception unit, and an interest analysis unit. The identity recognition unit is equipped with a 2-megapixel binocular camera and an RFID wristband that supports non-contact recognition within 10 meters, used to obtain user identity information; The behavior perception unit is equipped with an infrared human body sensor with a detection distance of 0.1-10m, a millimeter-wave radar with an identification accuracy of ±0.5m, and a touch screen interactive log collector to record user behavior data. The interest analysis unit is equipped with an NLP algorithm to parse user voice interaction content and extract interest keywords.

3. A media control system for an exhibition hall according to claim 1 or 2, characterized in that, The multi-dimensional sensing subsystem also includes an environmental data acquisition module and an equipment status acquisition module; The environmental data acquisition module is equipped with a temperature and humidity sensor with a measurement range of 0-60℃ and an accuracy of ±0.5℃, a light sensor with a measurement range of 0-100000 lux and an accuracy of ±5%, and a crowd density sensor with a counting error of ≤3%. The device status acquisition module is used to collect the power consumption, operating temperature and fault codes of the media device in real time, and upload them to the processing layer via the LoRa protocol with a transmission distance of ≤1km.

4. The media control system for an exhibition hall according to claim 1, characterized in that, The 5G+edge computing transmission subsystem includes a transmission network and edge nodes; The transmission network adopts a 5G private network with a downlink rate of ≥1Gbps, an uplink rate of ≥100Mbps and a latency of ≤10ms, and is configured with Ethernet backup; Each edge node is one edge server deployed in each exhibition area; The edge nodes can perform data preprocessing, local emergency control in the event of core network failure, and direct issuance of commands with high real-time requirements.

5. A media control system for an exhibition hall according to claim 1, characterized in that, The intelligent control core subsystem includes a user profile engine, which constructs a three-dimensional user profile based on data collected from the perception layer, including identity, behavior, and interests. It then uses a collaborative filtering algorithm that combines user similarity and content similarity to generate a personalized content recommendation list, and the update frequency of the recommendation list is ≤10 seconds.

6. A media control system for an exhibition hall according to claim 1 or 5, characterized in that, The intelligent control core subsystem also includes a dynamic resource scheduling module, which includes a server resource scheduling unit and a media resource scheduling unit. The server resource scheduling unit uses an LSTM model to predict the crowd density in each exhibition area over the next 30 minutes and dynamically adjusts the computing power allocation between the core server and edge nodes. During peak hours, the core server handles 70% of the computing power and the edge nodes handle 30%. During off-peak hours, the core server sleeps with 50% of the computing power and the edge nodes handle 50%. The media resource scheduling unit can establish a content cache pool that supports 4K / 8K video and 3D models. When the content is accessed ≥5 times in the past hour, the content is preloaded to the edge node.

7. A media control system for an exhibition hall according to claim 1, characterized in that, The intelligent control core subsystem also includes a cross-exhibition area linkage control module. This module tracks the user's location based on UWB positioning technology with a positioning accuracy of ±10cm. When the user moves from exhibition area A to exhibition area B, a content connection protocol is triggered, causing exhibition area B to play a sequel related to the content in exhibition area A. At the same time, a device collaboration protocol is triggered, causing the lights in exhibition area A to decrease from 100% to 30% within 5 seconds, and the lights in exhibition area B to increase from 30% to 100% within 5 seconds. The audio signal is also switched to the audio signal of exhibition area B.

8. A media control system for an exhibition hall according to claim 1, characterized in that, The low-carbon energy-saving engine can turn off 50% of the screens and keep one core screen playing the summary in a loop when the density of people in the exhibition area is ≤0.1 people / ㎡, while reducing the brightness of the lights to 50%; the low-carbon energy-saving engine can reduce the screen brightness to 70% and turn off the artificial lighting on the same side when the natural light is ≥5000 lux; the low-carbon energy-saving engine can enter a low-power mode with a power consumption of 20% of the normal mode when the device has no interaction time for ≥15 minutes, and wake up the device within 1 second when a user approaches to a distance of ≤2m.

9. A media control system for an exhibition hall according to claim 1, characterized in that, The content output module of the media output and interaction subsystem supports multi-terminal adaptation and multiple content formats. Multi-terminal compatibility includes LED splicing screens that support up to 100 screens, transparent OLED screens with a transparency of ≥40%, and AR glasses that support spatial anchoring and can overlay 3D virtual exhibits. Multiple content formats are available, including 4K / 8K video, 3D models that support gesture control for rotation / zoom, and VR panoramas with a 360° field of view.

10. A media control system for an exhibition hall according to claim 1, characterized in that, The user interaction module of the media output and interaction subsystem includes a multimodal interaction unit and a personalized customization unit; The multimodal interaction unit supports Chinese and English voice interaction, gesture interaction that recognizes six gestures (click, swipe, zoom, rotate, fist, and wave) through a TOF camera, and AR gesture grasping for virtual grabbing of exhibit models. The personalized customization unit allows users to save content via touchscreen. The system generates a unique QR code, which users can scan to obtain a content link valid for 72 hours.

Images