LED large screen synchronous control method, device and equipment of smart industrial city

Abstract

This application provides a method, apparatus, and equipment for synchronous control of LED screens in smart industrial cities, relating to the field of smart industrial city technology. The method involves acquiring the screen profile information and operating status information of each distributed LED screen; selecting multiple target data from multi-source data based on the screen profile information and operating status information, and fusing them to obtain clock-aligned fused data; performing causal correlation analysis on the clock-aligned fused data, and generating control commands based on causal logic; calculating the clock offset between the standard system time of the service control center and the local system time of each distributed LED screen, and calculating the network transmission delay between the central LED screen and each distributed LED screen; calculating the target playback trigger time for each distributed LED screen, encapsulating the target playback trigger time and control commands, and sending them to the corresponding distributed LED screens to trigger them to perform display operations, thereby achieving dynamic synchronous display of multiple LED screens.

Description

Technical Field

[0001] This application relates to the technical field of smart industrial cities, and specifically relates to a method, device, and equipment for synchronously controlling LED large screens in a smart industrial city. Background Art

[0002] In the process of building a smart industrial city, as a key visual terminal, LED large screens are widely used in scenarios such as the service control center of a smart industrial city, the monitoring hall of an industrial park, and public service areas, for displaying various information such as industrial production parameters, traffic flow, environmental monitoring data, and energy consumption. Currently, the operation data of a smart industrial city is scattered and stored in multiple independent systems, including the production data system of an industrial Internet platform, a city traffic management system, an environmental monitoring system, an energy management system, etc. The data formats of each system are significantly different, and there is a serious "data island" problem.

[0003] In related technologies, the control of LED large screens in the service control center mostly adopts a single data source access mode, or imports the data of multiple systems into the large screen control system after manual summarization. This not only has low efficiency but also has data latency. Moreover, the LED large screen control systems in each distributed area are independent and lack a unified spatio-temporal reference. When multi-screen联动展示 is required, the画面切换 is out of sync, reducing the intuitiveness of the display of the urban operation situation. Summary of the Invention

[0004] This application provides a method, device, and equipment for synchronously controlling LED large screens in a smart industrial city, which can achieve the efficient integration of multi-system data and the dynamic synchronous display of multiple LED large screens, improving the display effect.

[0005] The technical solution of the embodiment of this application is as follows: In a first aspect, the embodiment of this application provides a method for synchronously controlling LED large screens in a smart industrial city, which is applied to a service control center. The central LED large screen of the service control center communicates with the distributed LED large screens in multiple distributed areas of the smart industrial city. The method includes: Obtain the large screen portrait information and operating status information of each of the distributed LED large screens. The large screen portrait information includes spatial position information and attribute information, and the operating status information includes the local system time; Obtain multi-source data with spatio-temporal tags, screen out multiple target data matching each of the distributed LED large screens from the multi-source data according to the spatial position information and the attribute information, and perform spatio-temporal alignment and fusion processing on each of the target data based on the attribute information, the local system time, and the spatial position information to obtain fusion data with clock alignment; Perform causal association analysis on the clock-aligned fusion data to extract the linkage logic between the data, obtain the causal logic relationship, and generate a control instruction including the display time sequence and the screen switching logic according to the causal logic relationship; Obtain the standard system time of the service control center, calculate the clock offset value between the standard system time and each local system time, and calculate the network transmission delay between the central LED large screen and each distributed LED large screen; Based on the clock offset value and the corresponding network transmission delay, calculate the target playback trigger time of each distributed LED large screen, encapsulate the target playback trigger time and the control instruction, and send them to the corresponding distributed LED large screens respectively to trigger each distributed LED large screen to execute the display operation.

[0006] In the above technical solution, first, obtain the portrait information and operation status information of each distributed LED large screen, which provides a basis for subsequent spatial data matching and time synchronization calculation. Also, obtain multi-source data with spatio-temporal tags, and perform data screening, spatio-temporal alignment, and fusion processing to obtain clock-aligned fusion data, which not only realizes automatic import into the service control center but also makes each distributed LED large screen with different spatio-temporal benchmarks highly consistent in space and time, realizing the efficient integration of multi-system data and preparing for subsequent data processing. Perform causal association analysis on the clock-aligned fusion data to dynamically adjust the display content according to the actual situation of the industrial field, generate a control instruction including the display time sequence and the screen switching logic, and realize the linkage of each distributed LED large screen; calculate the clock offset value and network transmission delay, encapsulate the target playback trigger time and the control instruction, and send them to each distributed LED large screen to trigger each distributed LED large screen to synchronously execute the display operation, realizing the dynamic synchronous display of multiple LED large screens and improving the display effect.

[0007] In some embodiments of the present application, the performing causal association analysis on the clock-aligned fusion data to extract the linkage logic between the data and obtain the causal logic relationship includes: Input the clock-aligned fusion data into a preset spatio-temporal neural network model to extract a high-order feature vector sequence including spatial topological features and time-dependent features; Perform Granger causality test on the high-order feature vector sequence, identify the leading feature variables, and establish a candidate linkage event set according to the leading feature variables; Use the difference-in-differences method to verify the effectiveness of the candidate linkage event set and obtain the difference-in-differences effect value; In the case where the difference-in-differences effect value is greater than a preset net impact threshold, determine that there is a causal relationship between the leading feature variable and the display content in the operation status information, and obtain the causal logic relationship according to the causal relationship.

[0008] In some embodiments of this application, the step of using the difference-in-differences method to verify the effectiveness of the candidate linked event set and obtain the difference-in-differences effect value includes: For each of the leading feature variables in the candidate linkage event set, a processing data group containing the current real-time data sequence is constructed; Obtain data of the leading characteristic variable during historical normal periods and construct a control data set; Calculate the trend values ​​of the data processing group and the control data group within a preset observation window to obtain the first trend value corresponding to the data processing group and the second trend value corresponding to the control data group. The effect value is obtained by calculating the difference-in-differences effect value using the first trend value and the second trend value.

[0009] In some embodiments of this application, the spatiotemporal neural network model includes a spatial feature encoder, a temporal extraction module, a dual attention mechanism module, and a feature mapping module; The step of inputting the clock-aligned fused data into a preset spatiotemporal neural network model to extract a high-order feature vector sequence containing spatial topological features and time-dependent features includes: The spatial feature encoder is used to extract multi-scale spatial features from the clock-aligned fused data to obtain a sequence of spatial feature maps; The temporal extraction module is used to perform spatiotemporal memory transfer on the spatial feature map sequence to obtain a temporal feature map sequence; The spatial feature map sequence and the temporal feature map sequence are weighted and fused using the dual attention mechanism module to obtain a spatiotemporal feature map; The spatiotemporal feature map is reduced in dimensionality using the feature mapping module to obtain the high-order feature vector sequence.

[0010] In some embodiments of this application, the step of performing spatiotemporal alignment and fusion processing on each of the target data based on the attribute information, the local system time, and the spatial location information to obtain clock-aligned fused data includes: The spatial influence radius and effective time window of each of the distributed LED screens are determined based on the attribute information. A spatial filtering domain is constructed using the spatial location information as the center and the spatial influence radius as the threshold; simultaneously, a dynamic time filtering domain is constructed by mapping the effective time window using the local system time as the reference anchor point. The target data is mapped to the spatial filtering domain and the dynamic temporal filtering domain, and the spatiotemporal correlation score of the target data relative to the distributed LED screen is calculated. Select target data whose spatiotemporal correlation score is greater than a preset fusion threshold, and perform time-series mapping and splicing processing on the selected data to obtain the clock-aligned fused data.

[0011] In some embodiments of this application, mapping the target data to the spatial filtering domain and the dynamic temporal filtering domain, and calculating the spatiotemporal correlation score of the target data relative to the distributed LED screen, includes: In the spatial filtering domain, the Euclidean distance between the data location of the target data and the location of the large screen is calculated to obtain the distance value. The distance value is then weighted using a preset Gaussian decay function to obtain the spatial weight. In the dynamic time filtering domain, it is determined whether the generation time of the target data falls within the effective time window. If it is within the effective time window, the difference between the local system time and the generation time is calculated to obtain the time offset, and the time weight is calculated using the time offset. The spatial weight and the temporal weight are weighted and fused to obtain the spatiotemporal correlation score.

[0012] In some embodiments of this application, calculating the time weight using the time offset includes: Multiply the time offset by a preset adjustment parameter to obtain the adjustment result; The adjustment result is added to the preset first parameter, and then the reciprocal calculation is performed to obtain the time weight.

[0013] Secondly, embodiments of this application provide a synchronous control device for LED screens in a smart industrial city, applied in a service control center. The central LED screen of the service control center communicates with distributed LED screens in multiple areas of the smart industrial city. The device includes: The first acquisition module is used to acquire the screen image information and operating status information of each of the distributed LED screens. The screen image information includes spatial location information and attribute information, and the operating status information includes local system time. The second acquisition module is used to acquire multi-source data with spatiotemporal tags, filter out multiple target data that match each of the distributed LED screens from the multi-source data according to the spatial location information and the attribute information, and perform spatiotemporal alignment and fusion processing on each of the target data based on the attribute information, the local system time and the spatial location information to obtain clock-aligned fused data. The data analysis module is used to perform causal correlation analysis on the clock-aligned fused data to extract the linkage logic between the data, obtain the causal logic relationship, and generate control instructions containing display timing and screen switching logic based on the causal logic relationship. An acquisition calculation module, configured to acquire the standard system time of the service control center, calculate the clock offset value between the standard system time and each local system time, and calculate the network transmission delay between the central LED large screen and each distributed LED large screen; A synchronization distribution module, configured to calculate the target play trigger time of each distributed LED large screen based on the clock offset value and the corresponding network transmission delay, encapsulate the target play trigger time and the control instruction, and send them to the corresponding distributed LED large screens respectively to trigger each distributed LED large screen to execute a display operation.

[0014] In a third aspect, an embodiment of the present application provides an electronic device, including a processor, a memory, a user interface, a communication bus, and a network interface. The processor, the memory, the user interface, and the network interface are respectively connected to the communication bus. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the method according to any one of the first aspect.

[0015] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores instructions. When the instructions are executed, the method according to any one of the first aspect is executed.

[0016] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. By first acquiring the portrait information and operating status information of each distributed LED large screen, it provides a basis for subsequent spatial data matching and time synchronization calculation. It also acquires multi-source data with spatio-temporal tags, and performs data screening, spatio-temporal alignment, and fusion processing to obtain fusion data with clock alignment. This not only realizes automatic import into the service control center, but also makes each distributed LED large screen with different spatio-temporal benchmarks highly consistent in space and time, achieving efficient integration of multi-system data and preparing for subsequent data processing. Conduct causal correlation analysis on the fusion data with clock alignment to dynamically adjust the display content according to the actual situation of the industrial site, generate control instructions including display timing and screen switching logic, and realize the linkage of each distributed LED large screen; calculate the clock offset value and network transmission delay, encapsulate the target play trigger time and the control instruction, and send them to each distributed LED large screen to trigger each distributed LED large screen to synchronously execute a display operation, realizing dynamic synchronous display of multiple LED large screens and improving the display effect. Therefore, it effectively solves the problem of asynchronous multi-screen linkage display caused by single data access and lack of a unified spatio-temporal benchmark in the related art.

[0017] 2. By combining the spatiotemporal neural network model, Granger causality verification, and difference-in-differences verification, causal correlation analysis is performed on the clock-aligned fused data. This not only ensures the discovery of correlations but also removes spurious causal relationships, guaranteeing the accuracy of the linkage between various LED screens. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a method for synchronous control of LED screens in a smart industrial city, provided in one embodiment of this application. Figure 2 This is a schematic diagram of the connection relationship between various LED screens in a smart industrial city LED screen synchronous control method provided in one embodiment of this application; Figure 3 This is a schematic diagram of the spatiotemporal neural network model processing of the LED screen synchronization control method for smart industrial cities provided in another embodiment of this application; Figure 4 This is a schematic diagram of the module structure of a smart industrial city LED screen synchronous control device provided in one embodiment of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0019] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0020] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0021] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0022] In related technologies, existing service control center systems often have the following problems: First, the correlation between the displayed content and real-time data of the industrial site is low, making it impossible to intelligently respond to emergencies; second, the large screen control systems in each distributed area are independent and lack a unified spatiotemporal reference, resulting in asynchronous screen switching when multi-screen linkage display is required, which seriously affects the visual experience and command efficiency.

[0023] Based on this, embodiments of this application provide a method, device, electronic device, and readable storage medium for synchronous control of LED screens in smart industrial cities. The method first acquires the image information and operating status information of each distributed LED screen, providing a foundation for subsequent spatial data matching and time synchronization calculations. It also acquires multi-source data with spatiotemporal tags and performs data filtering, spatiotemporal alignment, and fusion processing to obtain clock-aligned fused data. This not only enables automatic integration into the service control center but also ensures high spatiotemporal consistency among distributed LED screens with different spatiotemporal references, achieving efficient integration of multi-system data and preparing for subsequent data processing. Causal correlation analysis is performed on the clock-aligned fused data to dynamically adjust the display content according to the actual situation in the industrial site, generating control commands containing display timing and screen switching logic to achieve linkage among the distributed LED screens. Furthermore, clock offset values ​​and network transmission delays are calculated, and the target playback trigger time and control commands are encapsulated and sent to each distributed LED screen, triggering synchronous display operations on each screen, achieving dynamic synchronous display of multiple LED screens and improving the display effect.

[0024] It should be noted that the LED screen synchronization control method of this smart industrial city can be used in the construction of smart industrial cities in various regions. By processing the data of each LED screen, it can achieve efficient integration of data from multiple systems and dynamic synchronous display of multiple LED screens, thereby improving the display effect.

[0025] The technical solutions provided in the embodiments of this application will be further described below with reference to the accompanying drawings.

[0026] Reference Figure 1 , Figure 1 This is a flowchart illustrating the LED screen synchronization control method for smart industrial cities provided in this application embodiment. The LED screen synchronization control method for smart industrial cities is applied in a service control center and is executed by a processor in an electronic device or a readable storage medium. The method includes steps S100, S200, S300, S400, and S500.

[0027] Step S100: Obtain the screen image information and operating status information of each distributed LED screen. The screen image information includes spatial location information and attribute information, and the operating status information includes local system time.

[0028] In one embodiment, each distributed LED screen can be a screen in an industrial production system or an urban operation and maintenance system. Attribute information includes the physical size, distribution screen identification parameters, resolution, and system affiliation of each distributed LED screen. Operating status information refers to the dynamic parameters of the screen at the current moment, including the local system time and displayed content. The displayed content is what the screen shows at the local system time, which is the current system clock reading of the control card inside the distributed LED screen. The service control center connects to the screens in each area via an industrial ring network. The center periodically issues query commands, and each distributed LED screen returns an XML configuration package to obtain its spatial location information, attribute information, and local system time. By obtaining the screen profile information and operating status information of each distributed LED screen, data support is provided for subsequent calculations.

[0029] like Figure 2 As shown, the central LED screen in the service control center communicates with distributed LED screens in various areas of the smart industrial city. These distributed LED screens include, but are not limited to, MES systems, SCADA systems, transportation systems, environmental protection systems, and energy systems. The screens in each of these systems are connected to the central LED screen. Data is transmitted through different protocols to obtain information from these distributed LED screens, providing data support for subsequent data integration and synchronized large-screen displays.

[0030] Step S200: Obtain multi-source data with spatiotemporal labels, filter out multiple target data matching each distributed LED screen from the multi-source data based on spatial location information and attribute information, and perform spatiotemporal alignment and fusion processing on each target data based on attribute information, local system time and spatial location information to obtain clock-aligned fused data.

[0031] In one embodiment, multi-source data refers to heterogeneous data aggregated in the data platform, including production work order progress from the MES system, equipment temperature / pressure values ​​from the SCADA system, GPS trajectories of logistics fleets from the transportation system, real-time pollution monitoring values ​​from the environmental protection system, and instantaneous voltage / current values ​​from the energy system. Target data is a dataset strongly correlated with a specific large screen after rule filtering, represented in the format {large screen ID, displayed content, time, location information}. For example, for a logistics intersection large screen, target data includes GPS data of vehicles waiting to enter. The service control center is configured with an adaptive access module. This module supports access for structured data (such as stored data tables) via ODBC / JDBC interfaces, unstructured data (such as displayed video content) via MQTT / HTTP protocols, and government data via REST API interfaces, achieving full coverage and seamless access to data from multiple systems. Automatic identification and connection of various data sources can be completed without manual intervention. The acquired data from each system is tagged with timestamps and geographic location labels. Heterogeneous data, carrying timestamps and geographic location information, is transmitted to the service control center, thereby obtaining multi-source data with spatiotemporal labels, providing a data foundation for subsequent data matching and filtering.

[0032] Then, each distributed LED screen is traversed. Based on spatial location and attribute information, data corresponding to the relevant attributes within that spatial location area is obtained, centered on the spatial location and with a preset distance as the radius. This data is then filtered from multi-source data to identify the relevant information for each distributed LED screen, extracting accurate data to improve the accuracy of subsequent calculations. The preset distance can be 1km.

[0033] In one embodiment, based on attribute information, local system time and spatial location information, spatiotemporal alignment and fusion processing is performed on each target data to obtain clock-aligned fused data, including but not limited to the following steps: Step S210: Determine the spatial influence radius and effective time window of each distributed LED screen based on the attribute information.

[0034] In one embodiment, a strategy mapping table is obtained. This table, stored in a hash table, records relevant information about each large screen, along with its corresponding spatial influence radius and effective time window. The relevant information includes attribute information, physical state information (on or off), and a summary of the displayed content. First, the large screen identifier of the distributed LED screens is obtained based on the attribute information. The corresponding large screen is then retrieved from the strategy mapping table using this identifier, and the identified distributed LED screen is determined based on its physical state. The corresponding spatial influence radius and effective time window are then directly retrieved from the identified distributed LED screen. This approach enables the acquisition of different spatiotemporal requirements for different types of large screens, avoiding the inability to identify differences between different large screens through uniform processing, and ensuring the real-time performance and effectiveness of different large screen scenarios. The strategy mapping table is created through statistical analysis of a large number of distributed LED screens, with professionals setting the spatial influence radius and effective time window based on the statistical analysis results.

[0035] For example, if the attribute information is for a SCADA alarm screen, the radius of the policy mapping table is 50m, and the effective time window is 10 seconds. If the attribute information is for a traffic overview screen, the radius of the policy mapping table is 5000m, and the effective time window is 30 minutes, etc.

[0036] Step S220: Construct a spatial filtering domain with spatial location information as the center and spatial influence radius as the threshold; simultaneously, construct a dynamic time filtering domain by mapping the effective time window with local system time as the reference anchor point.

[0037] In one embodiment, a spatial filtering domain is constructed using spatial location information as the center and the spatial influence radius as the threshold. This spatial filtering domain is a virtual geofence, typically a circular or polygonal geometric object, used to determine the spatial boundaries of the distributed LED screens, which is beneficial for subsequent calculations. A dynamic time filtering domain is constructed by mapping an effective time window using the local system time as the reference anchor point. This dynamic time filtering domain defines a closed interval of a sliding time window, within which the sliding time window is set according to different effective time windows. For example, if the local system time is represented as Tlocal and the effective time window is Window, then the start time is Tstart = Tlocal - Window, and the divided dynamic time filtering domain is [Tstart, Tlocal]. A sliding time step is set, which is based on the unit of the effective time window. If the unit is minutes, the sliding time step is per minute; if the unit is seconds, the sliding time step is per second. By constructing the dynamic time filtering domain as described above, even if there is network latency, it can be ensured that the filtered data is relative to the current time of the large screen, so as to obtain accurate fusion data and achieve synchronization between various large screens.

[0038] Step S230: Map the target data to the spatial filtering domain and the dynamic temporal filtering domain, and calculate the spatiotemporal correlation score of the target data relative to the distributed LED screen.

[0039] Specifically, the target data is mapped to a spatial filtering domain and a dynamic temporal filtering domain, and the spatiotemporal correlation score of the target data relative to the distributed LED screen is calculated, including but not limited to the following steps: Step S231: In the spatial filtering domain, calculate the Euclidean distance between the data location of the target data and the location of the large screen to obtain the distance value. Use a preset Gaussian decay function to calculate the weight of the distance value to obtain the spatial weight.

[0040] In some possible embodiments of this application, within the spatial filtering domain obtained in step S220, the Euclidean distance between the data location of the target data and the location of the large screen is calculated to obtain a distance value, which is then used to calculate the spatial weight. The calculation of the Euclidean distance is prior art and will not be elaborated upon here.

[0041] The Gaussian decay function is expressed as S= Substituting the distance values ​​obtained above into the Gaussian decay function, the closer to the large screen, the better. The smaller the value of , the larger the resulting S, meaning a larger spatial weight and a higher score in subsequent calculations. The Gaussian decay function allows distance values ​​to be processed through spatial weights as an intermediary, enabling subsequent calculations of spatiotemporal correlation scores. It's important to note that spatial weights calculate the spatial correlation between various data points to facilitate subsequent data fusion processing and improve the effectiveness of the fused data.

[0042] Step S232: In the dynamic time filtering domain, determine whether the generation time of the target data falls within the effective time window. If it is within the effective time window, calculate the difference between the local system time and the generation time to obtain the time offset, and use the time offset to calculate the time weight.

[0043] In some possible embodiments of this application, within the dynamic time filtering domain, it is determined whether the generation time of the target data falls within the effective time window. Specifically, it is determined whether the generation time is between the start time of the dynamic time filtering domain and the local system time. If it is between the start time and the local system time, it is within the effective time window; otherwise, it is not within the effective time window. If it is within the effective time window, the difference between the local system time and the generation time is calculated to obtain the time offset. This time offset is used to calculate the time weight, thereby calculating the spatiotemporal relevance score.

[0044] It should be noted that if the data is not within the valid time window, it indicates that the data acquisition corresponding to that time is unreasonable, and an alarm message is issued. After calibrating the local system time, step S100 is executed to obtain more accurate data and achieve accurate data integration between different systems.

[0045] In other possible embodiments of this application, the time weight is calculated using the time offset, including but not limited to: multiplying the time offset by a preset adjustment parameter to obtain an adjustment result; adding the adjustment result to a preset first parameter and then performing a reciprocal calculation to obtain the time weight.

[0046] Specifically, the time offset obtained above is denoted as Δt. The preset adjustment parameter k is set to 0.1, but it can also be set according to the update frequency of different distributed LED screens, which will not be elaborated here. First, the time offset is multiplied by the preset adjustment parameter to obtain the adjustment result. The adjustment parameter is used to adapt to the time scale of different distributed LED screens. Then, the adjustment result is added to the first parameter, and the reciprocal is calculated to obtain the time weight. An inverse proportional decay model is constructed. The closer the time, the smaller the time offset. By taking the reciprocal, the larger the result, the higher the weight, thus calculating a higher spatiotemporal correlation score. The first parameter is set to 1 so that the sum is greater than 1, thus constraining the reciprocal result to between 0 and 1 for easy calculation. It should be noted that the time weight calculates the degree of correlation between various data in time, so as to improve the effectiveness of data fusion processing in subsequent data fusion.

[0047] Step S233: The spatial weight and temporal weight are weighted and fused to obtain the spatiotemporal correlation score.

[0048] In some possible embodiments of this application, the first weighting coefficient and the second weighting coefficient of the spatial weight and temporal weight obtained above are acquired. The spatial weight is multiplied by the first weighting coefficient, and the temporal weight is multiplied by the second weighting coefficient. Finally, the results of the multiplications are added together to obtain the spatiotemporal correlation score. This spatiotemporal correlation score integrates spatial and temporal correlation data so that the data filtered based on the spatiotemporal correlation score is effective. The sum of the first weighting coefficient and the second weighting coefficient is 1. The ratio of the first weighting coefficient to the second weighting coefficient can be 6:4 or 7:3. Geographical location has a significant impact on the spatiotemporal correlation score, as spatial location information leads to different time offsets. Therefore, the ratio of the first weighting coefficient to the second weighting coefficient is set to be greater than that of the second weighting coefficient.

[0049] Step S240: Select target data with a spatiotemporal correlation score greater than a preset fusion threshold, and perform time-series mapping and splicing processing on the selected data to obtain clock-aligned fused data.

[0050] In one embodiment, the higher the spatiotemporal correlation score obtained from the above steps, the higher the correlation of the corresponding data. Target data with a spatiotemporal correlation score greater than a preset fusion threshold is selected to retain valid data, while target data with a spatiotemporal correlation score less than or equal to the fusion threshold is removed, improving data effectiveness. The preset fusion threshold can be 0.7. Then, the selected data undergoes time-series mapping, including resampling and interpolation completion, and the completed data is mapped to a unified standard time axis for time alignment. The time-aligned data is then concatenated to obtain clock-aligned fused data. This process addresses the issues of inconsistent data frequencies and time asynchrony across multiple industrial systems, ensuring that the data displayed on the large screen is consistent in spatiotemporal logic, achieving data integration, and facilitating subsequent synchronous control of various LED screens. Resampling and interpolation completion are existing technologies and will not be elaborated here. The data concatenation process specifically involves chaining the data together. The clock-aligned fused data is stored in JSON format for easy reading of the concatenated data, improving subsequent computational efficiency.

[0051] For example, since SCADA data is in the millisecond range (high frequency) and MES data is in the minute range (low frequency), the system uses the "local system time" on the large screen as the axis to perform hold-and-hold interpolation on the low-frequency MES data and downsample or mean aggregation on the high-frequency SCADA data, mapping them to the same time step (such as per second) to form a JSON fusion vector.

[0052] Step S300: Perform causal correlation analysis on the clock-aligned fused data to extract the linkage logic between the data, obtain the causal logic relationship, and generate control instructions containing display timing and screen switching logic based on the causal logic relationship.

[0053] In one embodiment, causal correlation analysis is performed on the clock-aligned fused data to extract the linkage logic between the data and obtain the causal logical relationship, including but not limited to the following steps: Step S310: Input the clock-aligned fused data into a preset spatiotemporal neural network model to extract a high-order feature vector sequence containing spatial topological features and time-dependent features.

[0054] In one embodiment, the higher-order feature vector is the hidden layer vector after model encoding, which removes noise and retains the essential features of the data. First, the clock-aligned fused data is converted into a data format to form a fused feature vector, which is represented as a multidimensional tensor of (B, T, N, C), where B is the batch size, T is the time step, N is the number of nodes, and C is the number of features. This fused feature vector is used as the input to the spatiotemporal neural network model.

[0055] like Figure 3As shown, the spatiotemporal neural network model includes a spatial feature encoder, a temporal extraction module, a dual attention mechanism module, and a feature mapping module. The spatial feature encoder employs a multi-scale spatial feature extraction module, including at least two different sets of convolutional kernels, used to extract local detail features and wide-area environmental features from the fused data, respectively. The temporal extraction module is a convolutional recurrent neural network used to extract temporal evolution patterns while maintaining the spatial topology of the fused data. The dual attention mechanism module is configured to calculate a spatial weight matrix and a temporal weight vector, adaptively weighting the output of the temporal extraction module to enhance the feature response values ​​of key spatiotemporal regions. The feature mapping module is a mapping module composed of nonlinear functions and fully connected layers, used to fuse and concatenate features to form a high-dimensional feature vector.

[0056] It should be noted that the spatiotemporal neural network model is a pre-trained model. A large amount of data from various large screens distributed throughout the city was crawled using a web crawler algorithm. This data was then cleaned and processed according to the data fusion method described above to obtain clock-aligned data. The processing steps are similar to the fusion process described above and will not be elaborated here. The processed data was used as training data to train the spatiotemporal neural network model. A regression loss function, such as mean squared error loss, was used. Then, reverse training was performed iteratively to obtain the pre-trained model. Finally, historical data from smart industrial cities was used to validate the pre-trained model, increasing its generalization ability and enabling accurate data processing using the spatiotemporal neural network model in subsequent applications.

[0057] In one embodiment, clock-aligned fused data is input into a preset spatiotemporal neural network model to extract a high-order feature vector sequence containing spatial topological features and temporal dependency features, including but not limited to the following steps: Step S311: Use a spatial feature encoder to extract multi-scale spatial features from the clock-aligned fused data to obtain a spatial feature map sequence.

[0058] Specifically, clock-aligned fused data is input into a spatial feature encoder for multi-scale spatial feature extraction, resulting in a sequence of spatial feature maps. The multi-scale spatial feature extraction employs two different sets of convolutional kernels: a first set of small-sized kernels (1×1 or 3×3) for extracting local features, with a small receptive field; and a second set of large-sized kernels (5×5 or 7×7) for extracting wide-area features, with a larger receptive field. The extracted features from the two sets are concatenated by channels, or reduced in dimensionality by a 1×1 convolutional layer, outputting the spatial feature map for each time step, denoted as {S1, S2...St}. This multi-scale architecture can both keenly capture local features and perceive large-scale, wide-area features of geographical locations, which is beneficial for determining whether there are correlations between the data from various distributed LED screens.

[0059] Step S312: Use the time sequence extraction module to perform spatiotemporal memory transfer on the spatial feature map sequence to obtain the time feature map sequence.

[0060] Specifically, the spatial feature map sequence extracted in step S311 is input into a convolutional recurrent neural network according to time steps for spatiotemporal memory transfer. This ensures that the spatial topological structure is not lost while processing the time series, resulting in a temporal feature map sequence, represented as {H1, H2...Ht}. Transferring memory through convolution ensures the transmission of spatiotemporal information, which is beneficial for analyzing the relationships between data.

[0061] Step S313: Use the dual attention mechanism module to perform weighted fusion of the spatial feature map sequence and the temporal feature map sequence to obtain the spatiotemporal feature map.

[0062] Specifically, the dual attention mechanism module configures a spatial weight matrix and a temporal weight matrix. The spatial weight matrix has a dimension of T×H×W×1. This dimension compresses channels through 1×1 convolutions and generates a spatial saliency mask using a sigmoid activation function. Each element represents the importance score of that location, with a score between 0 and 1. The temporal weight matrix has a dimension of T×1×1×C. This dimension utilizes global average pooling and a multilayer perceptron to generate channel response vectors, selecting the feature channels that evolve most dramatically over time. Each element represents the importance score of that historical moment. The spatial feature map sequence is multiplied by the spatial weight matrix, and the temporal feature map sequence is multiplied by the temporal weight matrix. The multiplication is performed element-wise, and then weighted and fused to obtain the spatiotemporal feature map. By directly fusing the lower-level spatial features with the higher-level temporal features, both trend judgment and accurate location positioning are ensured, significantly improving the capture of data correlations.

[0063] Step S314: Use the feature mapping module to perform dimensionality reduction mapping on the spatiotemporal feature map to obtain a high-order feature vector sequence.

[0064] Specifically, global average pooling is performed on the spatiotemporal feature map. The pooled features are then input into a nonlinear activation function to increase their expressive power, enabling them to fit nonlinear causal relationships. The nonlinear output features are then input into a fully connected layer to obtain a sequence of high-order feature vectors. For each distributed LED screen, each time step corresponds to a feature vector, which represents relevant information about that distributed LED screen and is also beneficial for subsequent causal verification.

[0065] Step S320: Perform Granger causality test on the high-order feature vector sequence to identify the leading feature variable, and establish a candidate linkage event set based on the leading feature variable.

[0066] Specifically, the Granger causality verification algorithm is used to perform causal verification on higher-order eigenvectors to identify leading feature variables. The Granger causality verification algorithm is existing technology and will not be elaborated upon here. Its causal verification algorithm solves for the leading feature variables that have Granger causes with the higher-order eigenvectors. An empty candidate linkage event set is created, and then each obtained leading feature variable is added to the candidate linkage event set to obtain the final candidate linkage event set. Given the large number of non-intuitive connections in smart industrial cities, the Granger test can automatically uncover these cross-system "butterfly effects" without the need for manually predefined rules. For example, traffic congestion leads to logistics delays, which in turn leads to raw material shortages in the MES system.

[0067] Step S330: The effectiveness of the candidate linkage event set is verified by using the difference-in-differences method to obtain the difference-in-differences effect value.

[0068] In one embodiment, the difference-in-differences method is used to verify the effectiveness of the candidate linkage event set and obtain the difference-in-differences effect value, including but not limited to the following steps: Step S331: For each leading feature variable in the candidate linkage event set, construct a processing data group containing the current real-time data sequence.

[0069] In some possible embodiments of this application, leading feature variables for a preset time period are extracted from the candidate linkage event set, wherein the preset time period can be the most recent 10 minutes. The extracted leading feature variables are arranged in chronological order to form a real-time data sequence, resulting in a processed data set, which provides a data foundation for subsequent verification.

[0070] Step S332: Obtain data of the leading characteristic variable during historical normal periods and construct a control data group.

[0071] In some possible embodiments of this application, historical data is obtained according to the preset time period in step S331, the data of the historical normal time period of the preset time period is extracted, the data of the historical normal time period is processed by data fusion according to the process of step S200, and the processed data is used to form a data sequence corresponding to the real-time data sequence according to the time series, and used as a control data group to provide a data basis for subsequent verification.

[0072] Step S333: Calculate the trend values ​​of the data processing group and the control data group within the preset observation window, respectively, to obtain the first trend value corresponding to the data processing group and the second trend value corresponding to the control data group.

[0073] In some possible embodiments of this application, for the data processing group, a first trend value is obtained by subtracting the data before the preset observation window from the data after the preset observation window. This first trend value reflects the changing trend of each distributed LED screen before and after the time. For the control data group, a second trend value is obtained by subtracting the data before the preset observation window from the data after the preset observation window. This second trend value reflects the changing trend of the historical normal data of each distributed LED screen before and after the time. The preset observation window corresponds to the effective time window.

[0074] Step S334: Calculate the effect value by combining the first trend value and the second trend value to obtain the difference-in-differences effect value.

[0075] In some possible embodiments of this application, the difference between the first trend value and the second trend value is calculated to obtain the difference-in-differences effect value. This difference reflects the reliability of the association calculated by the Granger causality verification algorithm; the larger the value, the more reliable the association, and the smaller the value, the less reliable the association.

[0076] Step S340: If the difference-in-differences effect value is greater than the preset net impact threshold, determine that there is a causal relationship between the leading feature variable and the content displayed in the running status information, and obtain the causal logical relationship based on the causal relationship.

[0077] In one embodiment, if the difference-in-differences effect value is greater than a preset net impact threshold, it indicates the existence of a correlation. This confirms a causal relationship between the leading feature variable and the displayed content in the operational status information. Interference from leading feature variables below the net impact threshold is removed, thus establishing connections between the various distributed LED screens. Based on these causal relationships, pointing edges are constructed to form causal logical relationships between the distributed LED screens in different systems.

[0078] In one embodiment, based on the generation time of the leading feature variable, the display time of each LED screen corresponding to that generation time is found. At that time node, the system's results triggered by the causal logic are displayed, forming the control logic for timing and screen switching. For example, if the traffic congestion occurs at time t1, the traffic system displays the traffic congestion situation at time t1. According to the causal logic, the traffic congestion causes logistics delays, so the logistics transportation system displays information about the logistics delays. This, in turn, leads to a raw material shortage in the MES system, so the MES system displays information about the lack of raw materials. This control instruction enables the distributed LED screens to work collaboratively to form the topology of a smart industrial city.

[0079] Step S400: Obtain the standard system time of the service control center, calculate the clock offset between the standard system time and each local system time, and calculate the network transmission delay between the central LED screen and each distributed LED screen.

[0080] In one embodiment, the standard system time is the precision atomic clock or GPS time synchronized by the service control center, serving as the global master clock. The difference between the standard system time and each local system time is calculated to obtain the clock offset value for each distributed LED screen, so that subsequent display control synchronization can be performed based on the clock offset value.

[0081] Network transmission latency is the one-way physical transmission time required for a data packet to travel from the central LED screen to the distributed LED screens. The network transmission delay is calculated using NTP or PTP algorithms, which are existing technologies and will not be elaborated upon here. The clock offset and network transmission latency are obtained above for subsequent synchronization control.

[0082] Step S500: Based on the clock offset value and the corresponding network transmission delay, calculate the target playback trigger time for each distributed LED screen, encapsulate the target playback trigger time and control command, and send them to the corresponding distributed LED screens to trigger each distributed LED screen to perform display operations.

[0083] In one embodiment, the target playback trigger time is an absolute future time point. All LED screens simultaneously refresh their pages at that moment, displaying the content corresponding to the control command. The service control center sets a reserved buffer time, which is greater than the maximum network latency among all LED screens. The target playback trigger time is obtained by adding the reserved buffer time to the standard system time. Then, for each distributed LED screen, its corresponding clock offset value is included in the target playback trigger time to form a trigger timestamp. The control command containing the unified trigger timestamp is encapsulated into a data packet, and the encapsulated data packet is sent to the corresponding distributed LED screens. Upon receiving the command, each distributed LED screen does not execute it immediately but places the control command in a queue and continuously compares it with its local system time. When the local system time minus the clock offset value equals the target playback trigger time, the screen is rendered immediately. The above process achieves synchronous display of all distributed LED screens, realizing intelligent control of smart industrial cities and improving the display effect.

[0084] like Figure 4As shown in the embodiment of this application, an LED screen synchronization control device for a smart industrial city is provided. This device is applied to a service control center, where the central LED screen communicates with distributed LED screens in multiple areas of the smart industrial city. The device acquires screen profile information and operating status information for each distributed LED screen through a first acquisition module 110. The screen profile information includes spatial location information and attribute information, and the operating status information includes local system time. A second acquisition module 120 acquires multi-source data with spatiotemporal tags. Based on the spatial location information and attribute information, multiple target data matching each distributed LED screen are selected from the multi-source data. Then, based on the attribute information, local system time, and spatial location information, spatiotemporal alignment and fusion processing are performed on each target data. The system obtains clock-aligned fused data; the data analysis module 130 performs causal correlation analysis on the clock-aligned fused data to extract the linkage logic between data and obtain causal logical relationships. Based on the causal logical relationships, control instructions containing display timing and screen switching logic are generated; the standard system time of the service control center is obtained through the acquisition and calculation module 140, the clock offset value between the standard system time and each local system time is calculated, and the network transmission delay between the central LED screen and each distributed LED screen is calculated; the synchronization distribution module 150 calculates the target playback trigger time of each distributed LED screen based on the clock offset value and the corresponding network transmission delay, encapsulates the target playback trigger time and control instructions, and sends them to the corresponding distributed LED screens to trigger each distributed LED screen to perform display operations.

[0085] It should be noted that the first acquisition module 110 is connected to the second acquisition module 120, the second acquisition module 120 is connected to the data analysis module 130, the data analysis module 130 is connected to the acquisition calculation module 140, and the acquisition calculation module 140 is connected to the synchronization distribution module 150. The aforementioned LED screen synchronization control method for smart industrial cities is applied to the LED screen synchronization control device 100 in smart industrial cities. The LED screen synchronization control device 100 acquires the image information and operating status information of each distributed LED screen, providing a foundation for subsequent spatial data matching and time synchronization calculation. It also acquires multi-source data with spatiotemporal tags and performs data filtering, spatiotemporal alignment, and fusion processing to obtain clock-aligned fused data. This not only enables automatic integration into the service control center but also ensures that the distributed LED screens with different spatiotemporal references are highly consistent in time and space, achieving efficient integration of multi-system data and preparing for subsequent data processing. Causal correlation analysis is performed on the clock-aligned fusion data to dynamically adjust the display content according to the actual situation in the industrial site, generate control commands containing display timing and screen switching logic to achieve linkage of various distributed LED screens; and calculate the clock offset value and network transmission delay, encapsulate the target playback trigger time and control commands and send them to each distributed LED screen to trigger each distributed LED screen to perform display operations synchronously, realize dynamic synchronous display of multiple LED screens, and improve the display effect.

[0086] It should also be noted that the apparatus provided in the above embodiments is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0087] This application also discloses an electronic device. (See reference...) Figure 5 , Figure 5 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. The electronic device 500 may include: at least one processor 501, at least one network interface 504, a user interface 503, a memory 505, and at least one communication bus 502.

[0088] The communication bus 502 is used to enable communication between these components.

[0089] The user interface 503 may include a display screen and a camera. Optionally, the user interface 503 may also include a standard wired interface and a wireless interface.

[0090] The network interface 504 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0091] The processor 501 may include one or more processing cores. The processor 501 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 505, and by calling data stored in memory 505. Optionally, the processor 501 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array. The processor 501 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and Modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 501.

[0092] The memory 505 may include random access memory (RAM) or read-only memory. Optionally, the memory 505 may include a non-transitory computer-readable storage medium. The memory 505 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 505 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 505 may also be at least one storage device located remotely from the aforementioned processor 501. (Refer to...) Figure 5 The memory 505, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a synchronous control method of LED screens in smart industrial cities.

[0093] exist Figure 5In the illustrated electronic device 500, the user interface 503 is mainly used to provide an input interface for the user and acquire user input data; while the processor 501 can be used to call an application program stored in the memory 505 for a synchronous control method of LED screens in a smart industrial city. When executed by one or more processors 501, the electronic device 500 performs one or more methods as described in the above embodiments. It should be noted that, for the foregoing method embodiments, for the sake of simplicity, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0094] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0095] In the various embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0096] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0097] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0098] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0099] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will readily conceive of those skilled in the art upon consideration of the specification and the disclosure of practical truths.

[0100] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for synchronous control of LED screens in smart industrial cities, characterized in that, The method is applied to a service control center, wherein the central LED screen of the service control center communicates with distributed LED screens in multiple areas of a smart industrial city, and the method includes: Obtain the screen image information and operating status information of each of the distributed LED screens. The screen image information includes spatial location information and attribute information. The operating status information includes local system time. Acquire multi-source data with spatiotemporal tags, filter out multiple target data matching each of the distributed LED screens from the multi-source data according to the spatial location information and the attribute information, and perform spatiotemporal alignment and fusion processing on each of the target data based on the attribute information, the local system time and the spatial location information to obtain clock-aligned fused data; A causal correlation analysis is performed on the clock-aligned fused data to extract the linkage logic between the data, obtain the causal logic relationship, and generate control instructions containing display timing and screen switching logic based on the causal logic relationship. Obtain the standard system time of the service control center, calculate the clock offset between the standard system time and each of the local system times, and calculate the network transmission delay between the central LED screen and each of the distributed LED screens; Based on the clock offset value and the corresponding network transmission delay, the target playback trigger time for each of the distributed LED screens is calculated. The target playback trigger time and the control command are then encapsulated and sent to the corresponding distributed LED screens to trigger each of the distributed LED screens to perform a display operation.

2. The method according to claim 1, characterized in that, The causal correlation analysis performed on the clock-aligned fused data to extract the linkage logic between the data and obtain the causal logical relationship includes: The clock-aligned fused data is input into a preset spatiotemporal neural network model to extract a high-order feature vector sequence containing spatial topological features and time-dependent features. Granger causality test is performed on the higher-order feature vector sequence to identify the leading feature variable, and a candidate linkage event set is established based on the leading feature variable; The effectiveness of the candidate linkage event set was verified using the difference-in-differences method to obtain the difference-in-differences effect value; If the difference-in-differences effect value is greater than the preset net impact threshold, it is determined that there is a causal relationship between the leading feature variable and the displayed content in the running status information, and the causal logical relationship is obtained based on the causal relationship.

3. The method according to claim 2, characterized in that, The validity verification of the candidate linked event set using the difference-in-differences method to obtain the difference-in-differences effect value includes: For each of the leading feature variables in the candidate linkage event set, a processing data group containing the current real-time data sequence is constructed; Obtain data of the leading characteristic variable during historical normal periods and construct a control data set; Calculate the trend values ​​of the data processing group and the control data group within a preset observation window to obtain the first trend value corresponding to the data processing group and the second trend value corresponding to the control data group. The effect value is obtained by calculating the difference-in-differences effect value using the first trend value and the second trend value.

4. The method according to claim 2, characterized in that, The spatiotemporal neural network model includes a spatial feature encoder, a temporal extraction module, a dual attention mechanism module, and a feature mapping module; The step of inputting the clock-aligned fused data into a preset spatiotemporal neural network model to extract a high-order feature vector sequence containing spatial topological features and time-dependent features includes: The spatial feature encoder is used to extract multi-scale spatial features from the clock-aligned fused data to obtain a sequence of spatial feature maps; The temporal extraction module is used to perform spatiotemporal memory transfer on the spatial feature map sequence to obtain a temporal feature map sequence; The spatial feature map sequence and the temporal feature map sequence are weighted and fused using the dual attention mechanism module to obtain a spatiotemporal feature map; The spatiotemporal feature map is reduced in dimensionality using the feature mapping module to obtain the high-order feature vector sequence.

5. The method according to claim 1, characterized in that, The process of performing spatiotemporal alignment and fusion processing on each of the target data based on the attribute information, the local system time, and the spatial location information to obtain clock-aligned fused data includes: The spatial influence radius and effective time window of each of the distributed LED screens are determined based on the attribute information. A spatial filtering domain is constructed using the spatial location information as the center and the spatial influence radius as the threshold; simultaneously, a dynamic time filtering domain is constructed by mapping the effective time window using the local system time as the reference anchor point. The target data is mapped to the spatial filtering domain and the dynamic temporal filtering domain, and the spatiotemporal correlation score of the target data relative to the distributed LED screen is calculated. Select target data whose spatiotemporal correlation score is greater than a preset fusion threshold, and perform time-series mapping and splicing processing on the selected data to obtain the clock-aligned fused data.

6. The method according to claim 5, characterized in that, The step of mapping the target data to the spatial filtering domain and the dynamic temporal filtering domain, and calculating the spatiotemporal correlation score of the target data relative to the distributed LED screen, includes: In the spatial filtering domain, the Euclidean distance between the data location of the target data and the location of the large screen is calculated to obtain the distance value. The distance value is then weighted using a preset Gaussian decay function to obtain the spatial weight. In the dynamic time filtering domain, it is determined whether the generation time of the target data falls within the effective time window. If it is within the effective time window, the difference between the local system time and the generation time is calculated to obtain the time offset, and the time weight is calculated using the time offset. The spatial weight and the temporal weight are weighted and fused to obtain the spatiotemporal correlation score.

7. The method according to claim 6, characterized in that, The calculation of the time weight using the time offset includes: Multiply the time offset by a preset adjustment parameter to obtain the adjustment result; The adjustment result is added to the preset first parameter, and then the reciprocal calculation is performed to obtain the time weight.

8. A synchronous control device for LED large screens in smart industrial cities, characterized in that, The device is applied to a service control center, where the central LED screen communicates with distributed LED screens in multiple areas of a smart industrial city. The device includes: The first acquisition module is used to acquire the screen image information and operating status information of each of the distributed LED screens. The screen image information includes spatial location information and attribute information, and the operating status information includes local system time. The second acquisition module is used to acquire multi-source data with spatiotemporal tags, filter out multiple target data that match each of the distributed LED screens from the multi-source data according to the spatial location information and the attribute information, and perform spatiotemporal alignment and fusion processing on each of the target data based on the attribute information, the local system time and the spatial location information to obtain clock-aligned fused data. The data analysis module is used to perform causal correlation analysis on the clock-aligned fused data to extract the linkage logic between the data, obtain the causal logic relationship, and generate control instructions containing display timing and screen switching logic based on the causal logic relationship. The calculation module is used to acquire the standard system time of the service control center, calculate the clock offset between the standard system time and each of the local system times, and calculate the network transmission delay between the central LED screen and each of the distributed LED screens. The synchronization distribution module is used to calculate the target playback trigger time of each of the distributed LED screens based on the clock offset value and the corresponding network transmission delay, and to encapsulate the target playback trigger time and the control command and send them to the corresponding distributed LED screens to trigger each of the distributed LED screens to perform display operations.

9. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, a communication bus, and a network interface. The processor, the memory, the user interface, and the network interface are respectively connected to the communication bus. The memory is used to store instructions. The user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1-7.

Images