Distributed intelligent operation and maintenance management method and system based on cloud computing

Abstract

The invention relates to the technical field of cloud computing operation and maintenance management, in particular to a distributed intelligent operation and maintenance management method and system based on cloud computing. And performing data processing on the divided data to obtain pre-processed data. And carrying out feature extraction on the preprocessed data. And performing distribution management on the extracted data based on an operation and maintenance management neural network. And obtaining a distributed management node data stream multi-dimensional graph of the distributed equipment for cloud computing. A huge data calculation processing program is decomposed into countless pieces of preprocessing data, and then the small programs are processed and analyzed through a system composed of a plurality of servers to obtain results and return the results to a user. Task distribution is solved, and calculation results are merged. And distributed calculation is effectively operated and maintained.

Description

Technical Field

[0001] This application relates to the technical field of cloud computing operation and maintenance management, and particularly to a distributed intelligent operation and maintenance management method and system based on cloud computing. Background Art

[0002] Cloud computing is a type of distributed computing, which refers to decomposing huge data computing programs into countless small programs through the network "cloud", and then processing and analyzing these small programs through a system composed of multiple servers to obtain results and return them to users. In the early days of cloud computing, simply put, it was simple distributed computing, which solved task distribution and merged calculation results. How to effectively perform operation and maintenance on distributed computing is a current industry difficulty. Therefore, it is necessary to propose a distributed intelligent operation and maintenance management method and system based on cloud computing for the defect that traditional distributed computing is difficult to effectively operate and maintain. Summary of the Invention

[0003] Based on this, it is necessary to propose a distributed intelligent operation and maintenance management method and system based on cloud computing for the defect that traditional distributed computing is difficult to effectively operate and maintain.

[0004] This application provides a distributed intelligent operation and maintenance management method based on cloud computing, including: Dividing the output data of distributed devices in cloud computing; Processing the divided data to obtain preprocessed data; Extracting features from the preprocessed data; Based on the operation and maintenance management neural network, performing distribution management on the extracted data; Obtaining a multi-dimensional graph of the distribution management node data stream of distributed devices in cloud computing.

[0005] Furthermore, establishing a binary digital value square matrix; Incorporating the output data into the binary digital value square matrix to obtain a numerical value square matrix; Cropping the numerical value square matrix; Obtaining a preprocessed value square matrix.

[0006] Furthermore, randomly screening the preprocessed value square matrix to form a training set, a validation set, and a test set.

[0007] Furthermore, generally enhancing the training set data; Performing standardization processing on the enhanced training set; Obtaining a preprocessed training set; Generally enhancing the validation set data; Performing standardization processing on the enhanced validation set; Obtain a preprocessed validation set; Perform general data augmentation on the test set; Perform standardization processing on the augmented test set; Obtain a preprocessed test set.

[0008] Furthermore, use the texture enhancement module TEM of the value matrix of the training set for grouping; Adjust the feature weights of the value matrix of the grouped training set; Assign attention weights to the value matrix of the training set to achieve feature extraction.

[0009] Furthermore, record the value matrix after feature extraction as a one-dimensional vector; Incorporate the one-dimensional vector into the global spatial average function; Use the global spatial average function to form the input layer of the operation and maintenance management neural network; Scale the vector of the input layer through the sigmoid function; Based on the residual structure of the operation and maintenance management neural network, achieve stable backpropagation of the gradient of the scaled vector.

[0010] Furthermore, fuse the vectors with stable backpropagation through the convolutional layer; Based on the channel dimension of the operation and maintenance management neural network, perform a multi-layer stacked convolutional layer; Obtain the ReLU activation function.

[0011] Furthermore, receive each ReLU activation function; Use Conv-BN-ReLU to obtain the foreground features of the data stream of the distribution management node; Optimize the foreground features based on the background attention component of the operation and maintenance management neural network; Obtain the multi-dimensional graph of the data stream of the distribution management node.

[0012] Furthermore, the operation and maintenance management neural network includes a feature extraction module and a texture enhancement module.

[0013] This application also provides a distributed intelligent operation and maintenance management system based on cloud computing, including: A server for executing the distributed intelligent operation and maintenance management method based on cloud computing; A memory communicatively connected to the server.

[0014] This application relates to a distributed intelligent operation and maintenance management method and system based on cloud computing. By dividing the output data of distributed devices in cloud computing, the divided data is processed to obtain preprocessed data. Feature extraction is performed on the preprocessed data. Based on the operation and maintenance management neural network, the extracted data is distributedly managed, and a multi-dimensional graph of the distributed management node data stream of the distributed devices in cloud computing is obtained. The huge data calculation and processing program is decomposed into countless preprocessed data, and then these small programs are processed and analyzed by a system composed of multiple servers to obtain results and return them to the user. Task distribution is solved, and the calculation results are merged. The distributed computing is effectively operated and maintained. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings, which form a part of this application, are used to provide a further understanding of this application, making other features, objectives, and advantages of this application more obvious. The schematic embodiments and descriptions of the accompanying drawings of this application are used to explain this application and do not constitute an improper limitation of this application.

[0016] Figure 1 It is a flowchart of a method for a distributed intelligent operation and maintenance management method based on cloud computing provided by an embodiment of this application.

[0017] Figure 2 It is a structural connection diagram of a distributed intelligent operation and maintenance management system based on cloud computing provided by an embodiment of this application.

[0018] Reference Signs: 100 - Server; 200 - Memory. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In order to make the objectives, technical solutions, and advantages of this application clearer, the following further elaborates on this application in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not used to limit this application.

[0020] This application provides a distributed intelligent operation and maintenance management method based on cloud computing.

[0021] As Figure 1 shown, in an embodiment of this application, a distributed intelligent operation and maintenance management method based on cloud computing includes: S100, dividing the output data of the distributed devices in cloud computing.

[0022] S200, processing the divided data to obtain preprocessed data.

[0023] S300, performing feature extraction on the preprocessed data.

[0024] S400 manages the distributed data based on the operation and maintenance management neural network for the extracted data.

[0025] S500 obtains the multi-dimensional graph of the distribution management node data stream of the distributed devices in cloud computing.

[0026] This embodiment relates to a distributed intelligent operation and maintenance management method and system based on cloud computing. By partitioning the output data of the distributed devices in cloud computing, the partitioned data is processed to obtain preprocessed data. Feature extraction is performed on the preprocessed data. Based on the operation and maintenance management neural network, the extracted data is managed distributively. The multi-dimensional graph of the distribution management node data stream of the distributed devices in cloud computing is obtained. The huge data calculation and processing program is decomposed into countless preprocessed data, and then these small programs are processed and analyzed by a system composed of multiple servers to obtain the results and return them to the user. It solves task distribution and merges the calculation results, effectively performing operation and maintenance on distributed computing.

[0027] In an embodiment of the present application, S100 includes: S111 establishes a binary digital value square matrix.

[0028] S112 incorporates the output data into the binary digital value square matrix to obtain a numerical value square matrix.

[0029] S113 trims the numerical value square matrix.

[0030] S114 obtains a preprocessed value square matrix.

[0031] Specifically, the data set is preprocessed, trimmed, and preprocessed to form a preprocessed data set, and the data set is divided into a training set, a validation set, and a test set.

[0032] The data set is subjected to general data augmentation or normalization processing to obtain a binary digital value square matrix. The output data is incorporated into the binary digital value square matrix to obtain a numerical value square matrix.

[0033] S121 randomly selects the preprocessed value square matrix to form a training set, a validation set, and a test set.

[0034] After obtaining the preprocessed value square matrix, the training set images of the data set can be input into the feature extraction module of the deep convolutional neural network in batches to extract multi-level features.

[0035] Specifically, the random selection can be implemented using a random sampling function, such as the RANSAC theory.

[0036] In an embodiment of the present application, S200 includes: S210 enhances the general data of the training set.

[0037] S220, performing standardization processing on the enhanced training set.

[0038] S230, obtaining a preprocessed training set.

[0039] S240, enhance the common data of the validation set.

[0040] S250, performing standardization processing on the enhanced validation set.

[0041] S260, obtain a preprocessed validation set.

[0042] S270, enhance the common data of the test set.

[0043] S280, performing standardization processing on the enhanced test set.

[0044] S290, obtain a preprocessed test set.

[0045] In one embodiment of the present application, S300 includes: S310, grouping using the texture enhancement module TEM of the value matrix of the training set.

[0046] S320, adjusting feature weights for the value matrix of the grouped training set.

[0047] S330, assigning attention weights to the value matrix of the training set to achieve feature extraction.

[0048] Specifically, for a convolutional feature map with a height of H, a width of W, and a number of channels of C, it is divided into K groups along the channel dimension, and each spatial position in each group of feature maps is represented as a one-dimensional vector, denoted as:

[0049] By applying the global spatial average function To approximate the semantic vector of the global features of this group learning representation .

[0050]

[0051] Using global semantic features TEM generates the corresponding importance coefficient for each local feature and performs a dot product operation to Evaluate the similarity between , then, through spatial normalization, the differences between different samples and the influence of noise are suppressed:

[0052]

[0053] in, and represent the mean and variance of the similarity coefficient respectively, represents a constant for numerical stability. To ensure that the normalization inserted into the network can represent the identity transformation, a pair of parameters are introduced for each coefficient and , and these parameters scale and shift the normalized values:

[0054] Finally, to obtain an enhanced texture feature vector , the original is scaled by the sigmoid function in the entire vector group space, and a residual structure is introduced to ensure stable backpropagation of the gradient:

[0055] All the enhanced features after being processed by the texture enhancement module form the final feature group:

[0056] By calculating the texture similarity at different spatial positions in the feature map and assigning higher weights to the spatial positions with similar texture characteristics, the texture enhancement module significantly improves the network's perception ability of the texture information of different ground objects in remote sensing images.

[0057] In an embodiment of the present application, S400 includes: S410, recording the value square matrix after feature extraction as a one-dimensional vector.

[0058] S420, incorporating the one-dimensional vector into the global spatial average function.

[0059] S430, using the global spatial average function to form the input layer of the operation and maintenance management neural network.

[0060] S440, scaling the vector of the input layer through the sigmoid function.

[0061] S450, based on the residual structure of the operation and maintenance management neural network, realizing stable backpropagation of the gradient of the scaled vector.

[0062] S460, fusing the vector with stable backpropagation in the convolutional layer.

[0063] S470, based on the channel dimension of the operation and maintenance management neural network, performing a multi-layer stacked convolutional layer.

[0064] S480, obtaining the ReLU activation function.

[0065] Specifically, the Sobel operator is applied in the horizontal and vertical directions to obtain the gradient map. First, two 3x3 parameters are initialized and a convolution operation with a step size of 1 is applied. The two convolution kernel parameters are initialized as Kx and Ky respectively.

[0066] These two convolutions are respectively applied to the input feature map to obtain the gradient map 、 、 。

[0067]

[0068]

[0069]

[0070] Among them, represents the convolution operation performed on the feature using a specific convolution kernel. The element sizes of the gradient maps and and respectively represent the boundary strengths of the feature in the horizontal direction and the vertical direction . represents the total boundary strength regardless of direction, which represents the vector sum of the horizontal gradient and the vertical gradient . After that, the gradient map is normalized by the function and fused with the input feature map to obtain the boundary-enhanced feature map .

[0071]

[0072] Among them, represents element-wise multiplication, and represents the function.

[0073] Fuses the boundary-enhanced feature maps of and with the simple stacked convolutional layers. Specifically, first, a bilinear upsampling operation and a 1x1 convolution are applied to the feature map to obtain a feature map with the same size as . After that, 1x1 convolution operations are respectively applied to adjust the channel sizes of these two features. Finally, these two feature maps are concatenated along the channel dimension, and a multi-layer stacked convolutional layer is applied to obtain the final boundary-enhanced feature and obtain it through the function :

[0074]

[0075]

[0076] Among them, represents an upsampling operation, represents a 1x1 convolution, represents a feature concatenation operation, represents function, represents passing through a 3x3 convolution, batch normalization, ReLU activation function, and 1x1 convolution in sequence.

[0077] In an embodiment of the present application, S500 includes: S510, receiving each ReLU activation function.

[0078] S520, obtaining the foreground features of the distribution management node data stream using Conv - BN - ReLU.

[0079] S530, optimizing the foreground features based on the background attention component of the operation and maintenance management neural network.

[0080] S540, obtaining the multi - dimensional graph of the distribution management node data stream.

[0081] Specifically, in the foreground path, by directly connecting and along the channel dimension, and then passing through Conv - BN - ReLU in sequence to obtain the foreground features :

[0082] Among them, represents element - wise multiplication, represents passing through Conv - BN - ReLU in sequence.

[0083] For the background path, the present invention designs a background attention component to enable the model to selectively focus on background information. The background features obtained by the background attention component are represented as:

[0084] Among them, represents a three - layer stacked Conv - BN - ReLU, represents the Sigmoid function, represents element - wise multiplication, Represents a background attention map, which is generated by applying the Sigmoid function to the feature map of the previous decoder layer and subtracting the resulting foreground attention map from 1.

[0085] The foreground features , the background features and the previous decoder features are concatenated in the channel dimension to form the output of the final decoder :

[0086] In an embodiment of the present application, the method further includes: S600, the operation and maintenance management neural network includes a feature extraction module and a texture enhancement module.

[0087] Specifically, the artificial neural network abstracts the human brain neuron network from the perspective of information processing, establishes a certain simple model, and forms different networks according to different connection methods. It may include a feature extraction module and a texture enhancement module.

[0088] A neural network is an operation model, based on a feature extraction module and a texture enhancement module, composed of a large number of nodes interconnected with each other. Each node represents a specific output function, called an activation function. The connection between each two nodes represents a weighted value for the signal passing through the connection, called a weight, which is equivalent to the memory of the artificial neural network. The output of the network varies depending on the connection method of the network, the weight value, and the activation function. And the network itself usually approximates a certain algorithm or function in nature, or may be an expression of a logical strategy.

[0089] In an artificial neural network, the neuron processing unit can represent different objects, such as features, letters, concepts, or some meaningful abstract patterns. The types of processing units in the network are divided into three categories: input units, output units, and hidden units. The input units receive signals and data from the external world. The output units implement the output of the system processing results. The hidden units are located between the input and output units and cannot be observed from the outside of the system. The connection weight values between neurons reflect the connection strength between units, and the representation and processing of information are reflected in the connection relationship of the network processing units. The artificial neural network is a non-programmed, adaptable, brain-style information processing, and its essence is to obtain a parallel distributed information processing function through the transformation and dynamic behavior of the network, and perform information processing functions at different levels and degrees.

[0090] An artificial neural network is a parallel distributed system that adopts a mechanism completely different from traditional artificial intelligence and information processing technologies. It overcomes the deficiencies of traditional logic symbol-based artificial intelligence in processing intuitive and unstructured information and has the characteristics of self-adaptation, self-organization, and real-time learning.

[0091] The distributed intelligent operation and maintenance management method based on cloud computing has virtualization. Virtualization breaks through the boundaries of time and space and is the most prominent feature of cloud computing. Virtualization technology includes application virtualization and resource virtualization. As we all know, there is no spatial connection between the physical platform and the application deployment environment. It is precisely through the virtual platform that operations on the corresponding terminals are completed for data backup, migration, and expansion, etc.

[0092] The distributed intelligent operation and maintenance management method based on cloud computing has high computing power. Adding cloud computing functions to the original server can rapidly increase the computing speed, and ultimately achieve the goal of dynamically expanding the virtualization level to expand the application.

[0093] The data resource libraries corresponding to the applications of the distributed intelligent operation and maintenance management method based on cloud computing are different. Therefore, when users run different applications, they need strong computing power to deploy resources, and the cloud computing platform can quickly allocate computing power and resources according to the needs of users.

[0094] The distributed intelligent operation and maintenance management method based on cloud computing all supports virtualization, such as storage networks, operating systems, and development software and hardware, etc. The virtualization elements are uniformly managed in the cloud system resource virtual pool. It can be seen that the compatibility of cloud computing is very strong. It can not only be compatible with low-configuration machines and hardware products of different manufacturers but also obtain higher-performance computing from peripherals.

[0095] If the server fails, it will not affect the normal operation of computing and applications. Because when a single-point server fails, the applications distributed on different physical servers can be restored through virtualization technology or new servers can be deployed using the dynamic expansion function for computing.

[0096] Managing resources in a unified virtual resource pool optimizes physical resources to a certain extent. Users no longer need expensive hosts with large storage spaces and can choose relatively inexpensive PCs to form a cloud. On the one hand, it reduces costs, and on the other hand, the computing performance is not inferior to that of large mainframes.

[0097] Users can utilize the rapid deployment conditions of the cloud computing-based distributed intelligent operation and maintenance management method to more simply and quickly expand their existing and new services as needed. For example, when a device fails in a computer cloud computing system, for users, there will be no hindrance either at the computer level or in specific applications. They can utilize the dynamic expansion function of computer cloud computing to effectively expand other servers. In this way, it can ensure that tasks are completed in an orderly manner. When dynamically expanding virtualized resources, applications can be efficiently expanded at the same time, improving the operation level of computer cloud computing.

[0098] This application provides a cloud computing-based distributed intelligent operation and maintenance management system.

[0099] Such as Figure 2 As shown, in an embodiment of this application, a cloud computing-based distributed intelligent operation and maintenance management system includes: The server 100 is used to execute the above-mentioned cloud computing-based distributed intelligent operation and maintenance management method.

[0100] The memory 200 is communicatively connected to the server 100.

[0101] This embodiment relates to a cloud computing-based distributed intelligent operation and maintenance management system. The server 100 divides the output data of the memory 200 of the distributed devices of cloud computing. The divided data is processed to obtain preprocessed data. Feature extraction is performed on the preprocessed data. Based on the operation and maintenance management neural network, the extracted data is distributedly managed. A multi-dimensional graph of the distributed management node data stream of the distributed devices of cloud computing is obtained. The huge data calculation processing program is decomposed into countless preprocessed data, and then these small programs are processed and analyzed by a system composed of multiple servers to obtain results and return them to the user. Task distribution is solved, and the calculation results are merged. The distributed computing is effectively operated and maintained.

[0102] The technical features of the above-described embodiments can be combined arbitrarily, and there is no limitation on the execution order of the method steps. For the sake of brevity of description, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as the combinations of these technical features do not conflict, they should all be considered as the scope described in this specification.

[0103] The above-described embodiments only represent several implementation manners of this application. Their descriptions are relatively specific and detailed, but they should not be construed as limiting the patent scope of this application. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several deformations and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of this application should be subject to the appended claims.

Claims

1. A distributed intelligent operation and maintenance management method based on cloud computing, characterized in that: include: Divide the output data of distributed devices in cloud computing; Processing the divided data to obtain preprocessed data; Perform feature extraction on preprocessed data; Based on the operation and maintenance management neural network, the extracted data is distributed and managed; A multi-dimensional graph of data flows of distributed management nodes of distributed devices in cloud computing is obtained.

2. The distributed intelligent operation and maintenance management method based on cloud computing according to claim 1 is characterized in that: The dividing of output data of the distributed devices of cloud computing includes: Create a binary digital value matrix; The output data is incorporated into a binary digital value matrix to obtain a numerical value matrix; Crop the numerical value matrix; Get the preprocessed value matrix.

3. The distributed intelligent operation and maintenance management method based on cloud computing according to claim 2 is characterized in that: The dividing of the output data of the distributed devices of cloud computing also includes: The preprocessed value matrix is ​​randomly screened to form training set, validation set and test set.

4. The distributed intelligent operation and maintenance management method based on cloud computing according to claim 3 is characterized in that: The step of processing the divided data to obtain preprocessed data includes: Enhance the training set with common data; Perform standardization on the enhanced training set; Get the preprocessed training set; Enhance the validation set with universal data; Perform standardization on the augmented validation set; Get the preprocessed validation set; Augment the test set with common data; Perform normalization on the augmented test set; Get the preprocessed test set.

5. The distributed intelligent operation and maintenance management method based on cloud computing according to claim 4 is characterized in that: The feature extraction of the preprocessed data comprises: The texture enhancement module TEM is used to group the value matrix of the training set; Adjust the feature weights for the value matrix of the grouped training set; Assign attention weights to the value matrix of the training set to achieve feature extraction.

6. The distributed intelligent operation and maintenance management method based on cloud computing according to claim 5 is characterized in that: The distributed management of the extracted data based on the operation and maintenance management neural network includes: The value matrix after feature extraction is recorded as a one-dimensional vector; Incorporate a one-dimensional vector into the global spatial average function; The global spatial average function is used to form the input layer of the operation and maintenance management neural network; Scale the vector of the input layer through the sigmoid function; Based on the residual structure of the operation and maintenance management neural network, stable backpropagation of the gradient of the scaled vector is achieved.

7. The distributed intelligent operation and maintenance management method based on cloud computing according to claim 6 is characterized in that: The distributed management of the extracted data based on the operation and maintenance management neural network also includes: The stable back-propagated vectors are fused into the convolutional layer; Based on the channel dimension of the operation and maintenance management neural network, multiple layers of stacked convolutional layers are performed; Get the ReLU activation function.

8. The distributed intelligent operation and maintenance management method based on cloud computing according to claim 7 is characterized in that: The step of obtaining a multi-dimensional graph of data flows of distributed management nodes of distributed devices of cloud computing includes: Receive each ReLU activation function; Use Conv-BN-ReLU to obtain the foreground features of the distributed management node data stream; Optimize foreground features based on the background attention component of the operation and maintenance management neural network; Obtain a multi-dimensional graph of data flows of distributed management nodes.

9. The distributed intelligent operation and maintenance management method based on cloud computing according to claim 8 is characterized in that: The method further comprises: The operation and maintenance management neural network includes a feature extraction module and a texture enhancement module.

10. A distributed intelligent operation and maintenance management system based on cloud computing, characterized in that: include: A server, configured to execute the distributed intelligent operation and maintenance management method based on cloud computing as described in any one of claims 1 to 9; The memory is communicatively connected with the server.

Images