Industrial data transmission system based on cloud computing

Abstract

The present application relates to the field of intelligent allocation, and specifically discloses an industrial data transmission system based on cloud computing, which uses probes to monitor and collect the amount of data to be transmitted in the industrial data transmission system in real time, and introduces data processing and analysis algorithms at the back end to perform timing analysis of the amount of data to be transmitted, so as to dynamically allocate and adjust computing resources according to the needs of data transmission, thereby improving the efficiency and quality of data transmission. In this way, dynamic optimization and management of computing resources in the data transmission process can be achieved, so as to better meet the needs of industrial data transmission, thereby promoting the development of the industrial Internet.

Description

Technical Field

[0001] The present application relates to the field of intelligent distribution, and more specifically, to an industrial data transmission system based on cloud computing. Background Art

[0002] Industrial data transmission system refers to the system used to transmit and manage data in the industrial field, covering various hardware and software components for collecting, processing, storing and transmitting data generated by industrial equipment and processes. In industrial data transmission systems, resource allocation refers to the reasonable allocation of limited resources (such as bandwidth, storage capacity, computing power, etc.) to different tasks or devices to meet system requirements and optimize system performance.

[0003] However, traditional industrial data transmission systems often adopt a static resource allocation method, that is, a fixed computing resource allocation plan is set based on experience or rules. However, this method cannot adapt to changes in data transmission needs and can easily lead to waste or shortage of computing resources, thereby affecting data transmission efficiency and quality.

[0004] Therefore, an optimized industrial data transmission system is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of the present application provides an industrial data transmission system based on cloud computing, which uses probes to monitor and collect the amount of data to be transmitted in the industrial data transmission system in real time, and introduces data processing and analysis algorithms at the back end to perform timing analysis of the amount of data to be transmitted, so as to dynamically allocate and adjust computing resources according to the needs of data transmission, thereby improving the efficiency and quality of data transmission. In this way, dynamic optimization and management of computing resources in the data transmission process can be achieved, so as to better meet the needs of industrial data transmission, thereby promoting the development of the industrial Internet.

[0006] According to one aspect of the present application, there is provided an industrial data transmission system based on cloud computing, comprising:

[0007] The data volume collection module to be transmitted is used to obtain the data volume to be transmitted at multiple predetermined time points within a predetermined time period collected by the probe;

[0008] A local time series dimension conversion module for data to be transmitted, used for performing local time series dimension conversion on the amount of data to be transmitted at the plurality of predetermined time points to obtain a sequence of local time series conversion images of the amount of data to be transmitted;

[0009] A module for extracting time series characteristics of the amount of data to be transmitted, used for extracting time series characteristics of the sequence of local time series conversion images of the amount of data to be transmitted to obtain a local time series correlation characteristic map of the amount of data to be transmitted;

[0010] A spatial saliency module, used for passing the local temporal correlation feature map of the amount of data to be transmitted through a spatial dimension saliency device based on a spatial self-attention layer to obtain a spatially significant local temporal correlation feature map of the amount of data to be transmitted as a spatially significant local temporal correlation feature of the amount of data to be transmitted;

[0011] The computing resource allocation module is used to determine whether the allocated computing resources should be increased, decreased or maintained based on the local temporal correlation characteristics of the spatially significant amount of data to be transmitted.

[0012] Compared with the prior art, the present application provides a cloud computing-based industrial data transmission system, which uses probes to monitor and collect the amount of data to be transmitted in the industrial data transmission system in real time, and introduces data processing and analysis algorithms at the back end to perform timing analysis of the amount of data to be transmitted, so as to dynamically allocate and adjust computing resources according to the needs of data transmission, thereby improving the efficiency and quality of data transmission. In this way, dynamic optimization and management of computing resources in the data transmission process can be achieved to better meet the needs of industrial data transmission, thereby promoting the development of the industrial Internet. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0014] Figure 1 It is a block diagram of an industrial data transmission system based on cloud computing according to an embodiment of the present application.

[0015] Figure 2 Schematic diagram of the architecture of an industrial data transmission system based on cloud computing according to an embodiment of the present application.

[0016] Figure 3 It is a block diagram of a local timing dimension conversion module for data to be transmitted in a cloud computing-based industrial data transmission system according to an embodiment of the present application.

[0017] Figure 4 It is a block diagram of a computing resource allocation module in an industrial data transmission system based on cloud computing according to an embodiment of the present application.

[0018] Figure 5 It is a block diagram of a feature optimization unit in an industrial data transmission system based on cloud computing according to an embodiment of the present application. DETAILED DESCRIPTION

[0019] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, but rather these embodiments are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0020] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0021] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0022] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0023] Industrial data transmission system refers to a system used to transmit and manage data in an industrial environment, usually composed of sensors and actuators, communication networks, data transmission protocols, data storage and processing systems, and monitoring and management systems. It can monitor and control industrial processes in real time, collect and analyze industrial data, thereby improving production efficiency and reducing costs, so as to support decision-making and optimization of industrial production and operations.

[0024] Resource allocation of industrial data transmission system refers to the reasonable configuration and management of various resources in the industrial data transmission system to meet the performance requirements and application requirements of the system. These resources include network bandwidth, computing resources, storage resources and sensors. Specifically, industrial data transmission systems usually need to transmit a large amount of data under limited network bandwidth. Resource allocation needs to consider the real-time and priority of data transmission, and allocate bandwidth to key data and applications to ensure their timely transmission and processing. In addition, industrial data transmission systems may need to perform real-time data processing, analysis and decision-making. Resource allocation needs to consider the complexity and priority of computing tasks, and allocate computing resources to key tasks and applications to meet real-time and performance requirements. In addition, industrial data transmission systems rely on sensors to collect real-time data. Resource allocation needs to consider the number, type and location of sensors, and allocate sensor resources to key monitoring points and applications to ensure the accuracy and integrity of data. Therefore, the strategies and methods of resource allocation can be adjusted and optimized according to the specific industrial data transmission system and application requirements. Through reasonable resource allocation, the performance, reliability and security of industrial data transmission systems can be improved to meet the needs of industrial applications.

[0025] However, traditional industrial data transmission systems usually adopt a static resource allocation method, which means that during the system design or deployment phase, a fixed computing resource allocation scheme is set based on experience or rules. This static resource allocation method may lead to the following problems. 1. Resource waste: Static resource allocation often determines resource configuration based on peak data transmission demand, which may lead to resource waste at low load. For example, if the system has only a small amount of data to be transmitted in a certain period of time, but static resource allocation configures computing resources to the level at peak load, it will cause a waste of computing resources. 2. Insufficient resources: Static resource allocation cannot adapt to the sudden increase in data transmission demand. If the system needs to process a large amount of data transmission in a certain period of time, but static resource allocation cannot provide enough computing resources, it will lead to reduced data transmission efficiency, increased latency, or even transmission failure. 3. Unable to cope with dynamic changes: Data transmission requirements in industrial environments often change dynamically. For example, the workload in different time periods may be different, and the data transmission priority of different applications may be different. Static resource allocation cannot be dynamically adjusted according to actual needs, and cannot flexibly respond to such changes, thus affecting resource utilization and data transmission efficiency.

[0026] Therefore, in response to the above technical problems, the technical concept of this application is to combine the industrial data transmission system with cloud computing, monitor and collect the amount of data to be transmitted in the industrial data transmission system in real time through probes, and introduce data processing and analysis algorithms at the back end to perform timing analysis of the amount of data to be transmitted, so as to dynamically allocate and adjust computing resources according to the needs of data transmission, thereby improving the efficiency and quality of data transmission. In this way, dynamic optimization and management of computing resources in the data transmission process can be achieved to better meet the needs of industrial data transmission, thereby promoting the development of the industrial Internet.

[0027] Figure 1 It is a block diagram of an industrial data transmission system based on cloud computing according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the architecture of an industrial data transmission system based on cloud computing according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to an embodiment of the present application, a cloud computing-based industrial data transmission system 100 includes: a data volume acquisition module 110 for acquiring the data volume to be transmitted at multiple predetermined time points within a predetermined time period collected by a probe; a local time series dimension conversion module 120 for data to be transmitted, for performing local time series dimension conversion on the data volume to be transmitted at the multiple predetermined time points to obtain a sequence of local time series conversion images of the data volume to be transmitted; a time series feature extraction module 130 for data volume to be transmitted, for performing time series feature extraction on the sequence of local time series conversion images of the data volume to be transmitted to obtain a local time series correlation feature map of the data volume to be transmitted; a spatial saliency module 140, for passing the local time series correlation feature map of the data volume to be transmitted through a spatial dimension saliency device based on a spatial self-attention layer to obtain a spatially significant local time series correlation feature map of the data volume to be transmitted as a spatially significant local time series correlation feature of the data volume to be transmitted; and a computing resource allocation module 150, for determining whether the allocated computing resources should be increased, decreased or maintained based on the spatially significant local time series correlation feature of the data volume to be transmitted.

[0028] In an embodiment of the present application, the module 110 for collecting the amount of data to be transmitted is used to obtain the amount of data to be transmitted at multiple predetermined time points within a predetermined time period collected by the probe. It should be understood that collecting the amount of data to be transmitted at multiple predetermined time points can help optimize resources. Specifically, the amount of data transmitted at different time points can reflect the dynamic changes in data transmission, including peaks and troughs, which is conducive to subsequent dynamic resource allocation according to actual needs. For example, in a time period with a low amount of transmitted data, the allocation of computing resources and network bandwidth can be reduced to save resources and improve resource utilization. In a time period with a high amount of transmitted data, resources can be increased to meet the needs of data transmission, thereby improving the overall transmission efficiency.

[0029] In an embodiment of the present application, the local time series dimension conversion module 120 of the data to be transmitted is used to perform local time series dimension conversion on the amount of data to be transmitted at the plurality of predetermined time points to obtain a sequence of local time series conversion images of the amount of data to be transmitted. It should be understood that considering that the time series image can more intuitively display the time series changes of the data, such as the fluctuation, trend and periodicity of the amount of data. Based on this, in the technical solution of the present application, by performing local time series dimension conversion on the amount of data to be transmitted at the plurality of predetermined time points to obtain a sequence of local time series conversion images of the amount of data to be transmitted, the law of data change can be better understood and more meaningful features can be extracted from the data. For example, a sliding window or time series analysis method can be used to convert the amount of data within a period of time into an image sequence, wherein each image represents the data situation within a time window. In this way, local features of the data, such as the rate of change of the data, the peak value, etc., as well as the relationship and trend between the data can be extracted, thereby providing a basis and support for the subsequent prediction of future data transmission trends.

[0030] Figure 3 : is a block diagram of a local time series dimension conversion module for data to be transmitted in a cloud computing-based industrial data transmission system according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 3 As shown, the local time series dimension conversion module 120 for data to be transmitted includes: a time series arrangement and segmentation unit 121 for data to be transmitted, which is used to arrange the data to be transmitted at the multiple predetermined time points into time series input vectors for data to be transmitted according to the time dimension, and then perform vector segmentation on the time series input vectors for data to be transmitted to obtain a sequence of local time series input vectors for data to be transmitted; and a vector-to-image conversion unit 122, which is used to pass the sequence of local time series input vectors for data to be transmitted through a vector-to-image converter to obtain a sequence of local time series conversion images for data to be transmitted.

[0031] Specifically, the time series arrangement and segmentation unit 121 of the amount of data to be transmitted is used to arrange the amount of data to be transmitted at the plurality of predetermined time points into a time series input vector of the amount of data to be transmitted according to the time dimension, and then perform vector segmentation on the time series input vector of the amount of data to be transmitted to obtain a sequence of local time series input vectors of the amount of data to be transmitted. It should be understood that, considering that industrial data often has a temporal correlation, that is, the amount of data to be transmitted at different time points has a certain temporal correlation relationship, and the amount of data to be transmitted has volatility and uncertainty in the time dimension, resulting in different time series patterns and change trends in different time periods in the time dimension. Therefore, in order to more fully and accurately capture the time series feature information of the amount of data to be transmitted, in the technical solution of the present application, after the amount of data to be transmitted at the plurality of predetermined time points is arranged into a time series input vector of the amount of data to be transmitted according to the time dimension, the time series input vector of the amount of data to be transmitted is vector segmented to obtain a sequence of local time series input vectors of the amount of data to be transmitted. By vectorizing the timing input vector of the amount of data to be transmitted, it is possible to subsequently capture the timing correlation and change pattern of the amount of data to be transmitted, so that the transmission process can be analyzed and monitored more accurately, and the transmission demand can be predicted to dynamically allocate and adjust computing resources.

[0032] Specifically, the vector-image conversion unit 122 is used to pass the sequence of the local time-series input vectors of the amount of data to be transmitted through a vector-image converter to obtain a sequence of local time-series conversion images of the amount of data to be transmitted. Accordingly, considering that compared with a simple time-series vector representation, time-series image data can provide more information, including the time-series relationship, fluctuation and trend changes of the amount of data to be transmitted. Therefore, in the technical solution of the present application, the sequence of the local time-series input vectors of the amount of data to be transmitted is further passed through a vector-image converter to obtain a sequence of local time-series conversion images of the amount of data to be transmitted. Such a conversion can better capture and analyze the time-series dynamic changes of the amount of data to be transmitted in each local time period. In other words, by converting the local time-series input vector of the amount of data to be transmitted into the form of an image, it is possible to easily extract the local time-series pattern and fluctuation characteristics of the amount of data to be transmitted in the local time-series conversion image of the amount of data to be transmitted, so that the time-series characteristics of the amount of data to be transmitted can be more intuitively observed and analyzed, which helps to more accurately analyze and predict the changes in data transmission demand.

[0033] In particular, in another achievable embodiment of the present application, the amount of data to be transmitted at the plurality of predetermined time points is converted into a local time series dimension to obtain a sequence of local time series conversion images of the amount of data to be transmitted, which can be achieved by the following steps. First, the amount of data to be transmitted at the plurality of predetermined time points is collected. This can be obtained by a sensor, a monitoring system or other data acquisition method. Then, the size of the time window is determined according to the requirements and the time series characteristics of the data. The time window represents the time range of the amount of data observed at each time point. For example, an hourly, daily or other suitable time window can be selected. Then, the amount of data to be transmitted is divided according to the time window, and for each time window, a time point is selected as the starting point of the window, and the amount of data to be transmitted in the window is obtained according to the size of the time window. Then, the amount of data to be transmitted in each time window is normalized. Normalization can map the amount of data to the same scale range to eliminate the order of magnitude difference between different time windows. Then, the normalized amount of data to be transmitted is converted into an image sequence. An image processing library or tool can be used to convert the amount of data in each time window into a corresponding image. For example, you can use pixel values ​​to represent the amount of data and map the amount of data to properties such as brightness or color of the image. After that, organize the converted images into an image sequence in time order. Make sure that the order of the image sequence is consistent with the order of the time window. Finally, visualize and analyze the image sequence. You can use image processing tools or data analysis tools to display the image sequence and perform further analysis. Observe the patterns, trends, and anomalies in the image sequence and make decisions and optimizations as needed.

[0034] In the embodiment of the present application, the temporal feature extraction module 130 of the amount of data to be transmitted is used to extract the temporal features of the sequence of the local temporal conversion images of the amount of data to be transmitted to obtain the local temporal correlation feature map of the amount of data to be transmitted. Accordingly, considering that the sequence of local temporal conversion images of the amount of data to be transmitted may contain a large amount of information, but not all information is important to the problem or task, it also contains important temporal change patterns and trends implicit in the amount of data to be transmitted. Based on this, in the technical solution of the present application, in order to extract more meaningful and relevant features from the sequence of local temporal conversion images of the amount of data to be transmitted, so as to extract these key information from the image sequence and understand the temporal characteristics and correlation of the data, the sequence of local temporal conversion images of the amount of data to be transmitted is further extracted to obtain the local temporal correlation feature map of the amount of data to be transmitted, so as to reveal the temporal relationship, correlation and dependency between the data, so as to better understand the temporal change law of the data, and better understand and analyze the temporal characteristics and correlation of the amount of data to be transmitted, and provide support for subsequent decision-making and optimization.

[0035] Specifically, in an embodiment of the present application, the time series feature extraction module of the amount of data to be transmitted is used to: pass the sequence of the local time series conversion images of the amount of data to be transmitted through the time series feature extractor of the amount of data to be transmitted based on the deep neural network model to obtain the local time series correlation feature map of the amount of data to be transmitted. In particular, the deep neural network model described here is a three-dimensional convolutional neural network model. That is, in order to be able to capture the change pattern of the amount of data to be transmitted in the time dimension and the correlation of the local time series, in the technical solution of the present application, the sequence of the local time series conversion images of the amount of data to be transmitted is further passed through the time series feature extractor of the amount of data to be transmitted based on the three-dimensional convolutional neural network model to perform feature mining to extract the local time series correlation feature information of the amount of data to be transmitted in each local time period, thereby obtaining the local time series correlation feature map of the amount of data to be transmitted. It is worth mentioning that the local time series correlation feature map of the amount of data to be transmitted can reflect the time series characteristics of the amount of data to be transmitted, such as trend, periodicity, mutation and volatility, so as to provide more accurate and effective input for subsequent processing steps.

[0036] It is worth mentioning that the temporal feature extractor of the amount of data to be transmitted based on the three-dimensional convolutional neural network model is a model for extracting features from a sequence of local temporal conversion images of the amount of data to be transmitted. It is an extension of the convolutional neural network (CNN) and is specifically designed to process three-dimensional data with temporal correlation, such as video, image sequence or time series data. Specifically, in a traditional two-dimensional convolutional neural network, the convolution operation is performed on the width and height dimensions of the image. In a three-dimensional convolutional neural network, the convolution operation also takes into account the time dimension, that is, the convolution operation is performed on the time dimension of the image sequence. This enables the three-dimensional convolutional neural network model to capture the temporal changes and correlations of the data in the time dimension. It should be understood that the temporal feature extractor of the amount of data to be transmitted based on the three-dimensional convolutional neural network is usually composed of multiple convolutional layers, pooling layers and fully connected layers. It can accept a sequence of local temporal conversion images of the amount of data to be transmitted as input, and extract temporal features through a series of convolution and pooling operations. These features can capture the patterns, trends and correlations of the data in the time series. That is to say, the time series feature extractor of the amount of data to be transmitted based on the three-dimensional convolutional neural network can accept the sequence of local time series conversion images of the amount of data to be transmitted as input. And use the three-dimensional convolution operation to extract features from the input data, and the convolution kernel slides in the width, height and time dimensions to extract local features. Then, the output of the convolution layer is nonlinearly mapped to increase the expression ability of the model and the feature map is downsampled through the three-dimensional pooling operation to reduce the size and parameter amount of the feature map while retaining important feature information. After that, the output of the pooling layer is flattened, and the feature fusion and classification are performed through the fully connected layer. Finally, according to the specific task requirements, the softmax function can be used for classification, the sigmoid function can be used for binary classification, or the linear activation function can be used for regression. Therefore, the time series feature extractor of the amount of data to be transmitted based on the three-dimensional convolutional neural network can extract time series correlation feature maps from the sequence of local time series conversion images of the amount of data to be transmitted, and these feature maps can be used for subsequent data analysis, pattern recognition, prediction and decision-making tasks. By utilizing the time series modeling capability of 3D CNN, the time series characteristics and correlation of data can be better understood, and the effect of data processing and analysis can be improved.

[0037] In an embodiment of the present application, the spatial saliency module 140 is used to pass the local temporal correlation feature map of the amount of data to be transmitted through a spatial dimension salient device based on a spatial self-attention layer to obtain a spatially salient local temporal correlation feature map of the amount of data to be transmitted as a spatially salient local temporal correlation feature of the amount of data to be transmitted. Accordingly, considering that in the temporal feature information of the amount of data to be transmitted, there will not only be key temporal feature information related to the data transmission requirements, but also some useless interference features. Therefore, in order to highlight the important spatial position information in the local temporal correlation feature map of the amount of data to be transmitted, so as to focus attention on the area that has a significant impact on the transmission process, so as to more accurately analyze and predict the data transmission requirements, so as to dynamically allocate computing resources, in the technical solution of the present application, the local temporal correlation feature map of the amount of data to be transmitted is further passed through a spatial dimension salient device based on a spatial self-attention layer to obtain a spatially salient local temporal correlation feature map of the amount of data to be transmitted. In particular, by inputting the local temporal correlation feature map of the amount of data to be transmitted into the spatial dimension salient device based on the spatial self-attention layer, the feature map can be spatially salient. Specifically, the spatial self-attention layer calculates the attention weight of each position based on the spatial relationship and interaction between features in the feature map. In this way, important areas will receive higher weights, while unimportant areas will receive lower weights, which can make the local temporal correlation feature map of the amount of data to be transmitted pay more attention to the key areas in the transmission process and reduce the interference of irrelevant or noise information. In this way, the expressive power of the feature map can be improved, and the important feature information in the temporal characteristics of the amount of data to be transmitted can be highlighted, providing more accurate and useful input for subsequent classification and analysis steps.

[0038] Specifically, in an embodiment of the present application, the spatial saliency module is used to: process the local temporal correlation feature map of the amount of data to be transmitted through a spatial dimension saliency device based on a spatial self-attention layer with the following spatial explicitation formula to obtain the spatially significant local temporal correlation feature map of the amount of data to be transmitted; wherein the spatial explicitation formula is:

[0039] F output =F input ⊙F self-attention

[0040] F self-attention =Active{Cov[Aug(F input )]}

[0041] Among them, F input is the local time series correlation characteristic diagram of the amount of data to be transmitted, F self-attentionis a spatial attention map, Active represents an activation function, Cov represents a convolutional layer, and Aug represents a low-cost augmentation of the local temporal correlation feature map of the amount of data to be transmitted when the size of the convolution kernel is greater than one, F output is the local temporal correlation feature map of the spatially significant data volume to be transmitted, and ⊙ represents the point multiplication by position.

[0042] In an embodiment of the present application, the computing resource allocation module 150 is used to determine whether the allocated computing resources should be increased, decreased, or maintained based on the spatially significant local time series correlation characteristics of the amount of data to be transmitted. Accordingly, it is considered that the spatially significant local time series correlation characteristics of the amount of data to be transmitted highlight the important spatial region information of the local time series correlation characteristics of the amount of data to be transmitted, and suppress the secondary areas. Based on this, in the technical solution of the present application, it is further determined based on the spatially significant local time series correlation characteristics of the amount of data to be transmitted that the allocated computing resources should be increased, decreased, or maintained, so that dynamic adjustments can be made according to the importance of different regions, and computing resources can be used more effectively. This can improve the utilization rate of computing resources and can better adapt to the needs of different regions, thereby improving the effect of data processing and analysis.

[0043] Figure 4 is a block diagram of a computing resource allocation module in an industrial data transmission system based on cloud computing according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 4 As shown, the computing resource allocation module 150 includes: a feature optimization unit 151, which is used to perform feature optimization on the spatially significant local time series correlation feature map of the amount of data to be transmitted to obtain an optimized spatially significant local time series correlation feature map of the amount of data to be transmitted; and a resource allocation judgment unit 152, which is used to pass the optimized spatially significant local time series correlation feature map of the amount of data to be transmitted through a classifier to obtain a classification result, and the classification result is used to indicate that the allocated computing resources should be increased, decreased or maintained.

[0044] It should be understood that in the technical solution of the present application, the local temporal correlation feature map of the amount of data to be transmitted represents the semantic correlation features of the sequence of local temporal conversion images of the amount of data to be transmitted based on the three-dimensional convolution kernel, that is, the contextual semantic correlation features of the local time domain pattern features of the amount of data to be transmitted in the time dimension. Considering that each local temporal conversion image of the amount of data to be transmitted in the sequence of local temporal conversion images of the amount of data to be transmitted has uneven distribution of source modal data along the time series, the semantic feature expression of each feature matrix of the local temporal correlation feature map of the amount of data to be transmitted along the channel dimension will be significantly unbalanced, which will be more significant after the local feature spatial distribution of the semantic features is strengthened by the spatial dimension saliency based on the spatial self-attention layer, thereby affecting the integrity of the feature distribution of the spatially significant local temporal correlation feature map of the amount of data to be transmitted as a whole, and further affecting the accuracy of the classification result obtained by the classifier of the spatially significant local temporal correlation feature map of the amount of data to be transmitted. Therefore, in the technical solution of the present application, the feature optimization is further performed on the local temporal correlation feature map of the spatially significant data volume to be transmitted to obtain an optimized local temporal correlation feature map of the spatially significant data volume to be transmitted.

[0045] Figure 5 is a block diagram of a feature optimization unit in an industrial data transmission system based on cloud computing according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 5 As shown, the feature optimization unit 151 includes: a linear transformation subunit 1511, which is used to perform a linear transformation on the spatially significant local temporal correlation feature map of the amount of data to be transmitted so that the width and height of each feature matrix along the channel dimension in the spatially significant local temporal correlation feature map of the amount of data to be transmitted are equal to obtain the converted spatially significant local temporal correlation feature map of the amount of data to be transmitted; and a feature optimization subunit 1512 of the amount of data to be transmitted, which is used to perform feature matrix-based optimization on each feature matrix along the channel dimension in the converted spatially significant local temporal correlation feature map of the amount of data to be transmitted to obtain the optimized spatially significant local temporal correlation feature map of the amount of data to be transmitted.

[0046] More specifically, in an embodiment of the present application, the feature optimization subunit of the amount of data to be transmitted is used to optimize each feature matrix along the channel dimension in the converted spatially significant local temporal correlation feature map of the amount of data to be transmitted based on the feature matrix to obtain the optimized spatially significant local temporal correlation feature map of the amount of data to be transmitted; wherein the optimization formula is:

[0047]

[0048] Among them, M i and M i+1are the feature matrices of the i-th and i+1-th positions of the local temporal correlation feature map of the spatially significant data to be transmitted after the conversion, and ε is the scale adjustment hyperparameter, M i+1 ′ is the characteristic matrix of the i+1th position along the channel direction of the local temporal correlation characteristic graph of the amount of data to be transmitted in the optimization space, represents matrix multiplication, ⊙ represents point multiplication by position, Represents matrix addition.

[0049] In particular, here, the coupled distribution direction of the local feature distribution of the converted spatially significant local temporal correlation feature map of the amount of data to be transmitted along the channel is predicted in the high-dimensional feature space through the progressive structured embedding calculation of the feature matrix with channel adjacent distribution of the converted spatially significant local temporal correlation feature map of the amount of data to be transmitted, so as to determine the iteratively generated transfer diagram representation based on channel coupling based on the progressive center of the distribution, and to reconstruct the contextual relationship of the converted spatially significant local temporal correlation feature map of the amount of data to be transmitted based on the scene layout of the feature matrix by refining the projection normalization proposal of the converted spatially significant local temporal correlation feature map of the amount of data to be transmitted as a whole from bottom to top along the channel dimension, so as to improve the integrity of the feature representation of the converted spatially significant local temporal correlation feature map of the amount of data to be transmitted, so as to improve the accuracy of the classification result obtained by the classifier. In this way, computing resources can be dynamically allocated and adjusted according to the needs of data transmission, thereby improving the efficiency and quality of data transmission. In this way, dynamic optimization and management of computing resources in the data transmission process can be achieved to better meet the needs of industrial data transmission and promote the development of the Industrial Internet.

[0050] Specifically, the resource allocation judgment unit 152 is used to pass the optimized spatially significant local time series associated feature map of the amount of data to be transmitted through a classifier to obtain a classification result, and the classification result is used to indicate whether the allocated computing resources should be increased, decreased, or maintained. That is, the key time series semantic feature information about the amount of data to be transmitted after spatial significant processing is used for classification processing to predict the data transmission demand, so as to dynamically allocate and adjust computing resources according to the data transmission demand to improve the efficiency and quality of data transmission. It should be understood that the classifier can evaluate the feature map to determine the importance of different regions and the degree of contribution to the classification task. Therefore, through the output results of the classifier, it can be determined which regions' features are critical to the classification task and which regions' features have less impact on the classification task. In this way, it can be determined whether the allocated computing resources should be increased, decreased, or maintained according to the importance of the features, thereby improving the utilization efficiency of computing resources.

[0051] Specifically, in an embodiment of the present application, the resource allocation judgment unit includes: a feature expansion unit, which is used to expand the optimization space significant local temporal correlation feature map of the amount of data to be transmitted into an optimization space significant local temporal correlation feature vector of the amount of data to be transmitted; a feature fully connected encoding unit, which is used to use the fully connected layer of the classifier to fully connect the optimization space significant local temporal correlation feature vector to obtain the optimization space significant local temporal correlation encoding feature vector; and a classification result generation unit, which is used to pass the optimization space significant local temporal correlation encoding feature vector of the amount of data to be transmitted through the Softmax classification function of the classifier to obtain the classification result.

[0052] In summary, the cloud computing-based industrial data transmission system 100 according to the embodiment of the present application is explained, which uses probes to monitor and collect the amount of data to be transmitted in the industrial data transmission system in real time, and introduces data processing and analysis algorithms at the back end to perform timing analysis of the amount of data to be transmitted, so as to dynamically allocate and adjust computing resources according to the needs of data transmission, thereby improving the efficiency and quality of data transmission. In this way, dynamic optimization and management of computing resources in the data transmission process can be achieved to better meet the needs of industrial data transmission, thereby promoting the development of the industrial Internet.

[0053] As described above, the cloud computing-based industrial data transmission system 100 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a cloud computing-based industrial data transmission algorithm. In a possible implementation, the cloud computing-based industrial data transmission system 100 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the cloud computing-based industrial data transmission system 100 can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the cloud computing-based industrial data transmission system 100 can also be one of the many hardware modules of the wireless terminal.

[0054] Alternatively, in another example, the cloud computing-based industrial data transmission system 100 and the wireless terminal may also be separate devices, and the cloud computing-based industrial data transmission system 100 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0055] The above descriptions of various implementations of the present disclosure are exemplary and not exhaustive. The disclosure is not limited to the disclosed implementations, and many modifications and variations are obvious to those of ordinary skill in the art without departing from the scope and spirit of the described implementations. The terms used herein are selected to best explain the principles of the implementations, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the various practical methods disclosed herein.

Claims

1. An industrial data transmission system based on cloud computing, characterized in that: include: The data volume collection module to be transmitted is used to obtain the data volume to be transmitted at multiple predetermined time points within a predetermined time period collected by the probe; A local time series dimension conversion module for data to be transmitted, used for performing local time series dimension conversion on the amount of data to be transmitted at the plurality of predetermined time points to obtain a sequence of local time series conversion images of the amount of data to be transmitted; A module for extracting time series characteristics of the amount of data to be transmitted, used for extracting time series characteristics of the sequence of local time series conversion images of the amount of data to be transmitted to obtain a local time series correlation characteristic map of the amount of data to be transmitted; A spatial saliency module, used for passing the local temporal correlation feature map of the amount of data to be transmitted through a spatial dimension saliency device based on a spatial self-attention layer to obtain a spatially significant local temporal correlation feature map of the amount of data to be transmitted as a spatially significant local temporal correlation feature of the amount of data to be transmitted; A computing resource allocation module, used to determine whether the allocated computing resources should be increased, decreased or maintained based on the local temporal correlation characteristics of the spatially significant amount of data to be transmitted; The spatial saliency module is used to: process the local temporal correlation feature map of the amount of data to be transmitted through a spatial dimension saliency device based on a spatial self-attention layer using the following spatial explicitation formula to obtain the spatially salient local temporal correlation feature map of the amount of data to be transmitted; The spatial manifestation formula is: ; in, is a local time series correlation characteristic graph of the amount of data to be transmitted, is the spatial attention map, represents the activation function, represents a convolutional layer, and It indicates that when the size of the convolution kernel is greater than one, a cheap augmentation is performed on the local temporal correlation feature map of the amount of data to be transmitted, The local temporal correlation feature map of the spatially significant amount of data to be transmitted is Represents point product by position.

2. The cloud computing-based industrial data transmission system according to claim 1, characterized in that: The local time series dimension conversion module of the data to be transmitted includes: a time series arrangement and segmentation unit for the amount of data to be transmitted, configured to arrange the amounts of data to be transmitted at the plurality of predetermined time points into time series input vectors of the amount of data to be transmitted according to the time dimension, and then perform vector segmentation on the time series input vectors of the amount of data to be transmitted to obtain a sequence of local time series input vectors of the amount of data to be transmitted; The vector-to-image conversion unit is used for converting the sequence of local time-series input vectors of the amount of data to be transmitted through a vector-to-image converter to obtain a sequence of local time-series conversion images of the amount of data to be transmitted.

3. The cloud computing-based industrial data transmission system according to claim 2, characterized in that: The module for extracting the time series features of the amount of data to be transmitted is used to: pass the sequence of the local time series conversion images of the amount of data to be transmitted through a time series feature extractor of the amount of data to be transmitted based on a deep neural network model to obtain the local time series correlation feature map of the amount of data to be transmitted.

4. The cloud computing-based industrial data transmission system according to claim 3, characterized in that: The deep neural network model is a three-dimensional convolutional neural network model.

5. The cloud computing-based industrial data transmission system according to claim 4, characterized in that: The computing resource allocation module comprises: A feature optimization unit, used for performing feature optimization on the spatially significant local time series correlation feature map of the amount of data to be transmitted to obtain an optimized spatially significant local time series correlation feature map of the amount of data to be transmitted; The resource allocation judgment unit is used to pass the local time series correlation feature map of the amount of data to be transmitted in the optimization space through a classifier to obtain a classification result, and the classification result is used to indicate whether the allocated computing resources should be increased, decreased or maintained.

6. The cloud computing-based industrial data transmission system according to claim 5, characterized in that: The feature optimization unit comprises: A linear transformation subunit, used for performing a linear transformation on the spatially significant local temporal correlation feature map of the amount of data to be transmitted so that the width and height of each feature matrix along the channel dimension in the spatially significant local temporal correlation feature map of the amount of data to be transmitted are equal to obtain a transformed spatially significant local temporal correlation feature map of the amount of data to be transmitted; The feature optimization subunit of the amount of data to be transmitted is used to optimize the various feature matrices along the channel dimension in the converted spatially significant local time series correlation feature map of the amount of data to be transmitted based on the feature matrix to obtain the optimized spatially significant local time series correlation feature map of the amount of data to be transmitted.

7. The cloud computing-based industrial data transmission system according to claim 6, characterized in that: The resource allocation judgment unit includes: A feature expansion unit, used for expanding the optimization space significant amount of data to be transmitted local time series correlation feature map into an optimization space significant amount of data to be transmitted local time series correlation feature vector; A feature fully connected encoding unit, used for using the fully connected layer of the classifier to perform fully connected encoding on the optimized spatially significant amount of data to be transmitted local temporal series associated feature vector to obtain an optimized spatially significant amount of data to be transmitted local temporal series associated encoded feature vector; The classification result generating unit is used to pass the local temporal correlation coding feature vector of the optimized spatially significant amount of data to be transmitted through the Softmax classification function of the classifier to obtain the classification result.

Images