Intelligent management method and system for application products of power dispatching edge cluster based on deep learning, medium and processor

Abstract

The invention discloses an intelligent management method and system for application products of a power dispatching edge cluster based on deep learning, a medium and a processor, and relates to the technical field of application management of computer software products, and the method comprises the steps: obtaining static features and dynamic features of the application products extracted from the power dispatching edge cluster; extracting first classification feature information of the static features and second classification feature information of the dynamic features through a neural network; splicing and fusing the first classification feature information and the second classification feature information, and outputting a category probability result after application rack product fusion in a trained classifier model; and matching a deployment strategy corresponding to the application product in a deployment strategy template library according to the category probability result. The static and dynamic characteristics of the application product are extracted, the neural network model is used for classification and prediction, a reasonable deployment strategy is finally generated, intelligent management of application product deployment is achieved, and the reasonability of system function module deployment is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of computer software product application management, and in particular to a method, system, medium and processor for intelligent management of application products of an electric power dispatching edge cluster based on deep learning. Background Art

[0002] As a computing paradigm that pushes data processing and storage to the edge of the network, edge computing has significant advantages such as low latency, high bandwidth, and reduced network congestion. Therefore, edge computing has gradually become an important way to solve the needs of big data processing.

[0003] With the expansion of edge clusters and the diversification of application scenarios, the management of application products has become more and more complicated. Traditional manual management methods have been unable to cope with this complexity. Therefore, a more intelligent and automated management method is needed to meet this challenge. For example, the application deployment method, device, terminal device and storage medium provided by China's public invention application CN117472390A record: According to the application product package, the application is released and the application instance is obtained. According to the application application request, the database is defined to obtain the database architecture; according to the database architecture, the microservice is defined to obtain the application instance information; the model is constructed through the database architecture and the application instance information to obtain the application model. According to the application model, code development is performed to obtain the development results; the development results, database architecture and application instance information are packaged to obtain the application product package. According to the database architecture, the data sub-library information is used to perform sub-library processing to obtain the data sub-library; according to the data sub-library, the application instance is deployed through the application instance information to obtain the application instance. According to the application application request obtained in advance, the application management platform deploys the application, obtains the application instance and receives the data routing rules corresponding to the application instance issued by the application management platform; the application deployment is realized and the efficiency of application deployment is improved. The prior art only proposes a method for realizing application deployment and accelerating application deployment, but does not provide how to realize intelligent management of application product deployment and deployment rationality.

[0004] In view of this, there is a need for an intelligent management method, system, medium and processor for application products of a power dispatching edge cluster based on deep learning. Summary of the invention

[0005] In view of the fact that the management of application products in the prior art is becoming more and more complicated, and the traditional management method is difficult to solve the complexity problem of application products, the present invention provides an application product intelligent management method, system, medium and processor of the power dispatch edge cluster based on deep learning, which can realize the intelligent management of application product deployment and improve the rationality of system function module deployment. The specific technical solution is as follows:

[0006] The present invention provides an intelligent management method for application products of a power dispatching edge cluster based on deep learning, comprising:

[0007] Obtain static and dynamic features of application products extracted from the power dispatch edge cluster;

[0008] Extracting first classification feature information of the static feature and second classification feature information of the dynamic feature through a neural network;

[0009] The first classification feature information and the second classification feature information are concatenated and fused, and the category probability result after the fusion of the application shelf products is output in the trained classifier model;

[0010] According to the category probability results, the deployment strategy corresponding to the application product is matched in the deployment strategy template library.

[0011] Preferably, the first classification feature information of the static feature extracted by using a neural network includes:

[0012] Preprocessing the data of the static features;

[0013] The processed static features are converted from high-dimensional sparse features of text types to low-dimensional dense vectors of numbers through an embedding layer, and the first classification feature information of the static features is extracted through a convolutional neural network.

[0014] Preferably, the second classification feature information of the dynamic feature extracted by using a neural network includes:

[0015] Preprocessing the data of the dynamic feature to convert the dynamic feature into a vector of fixed length;

[0016] The second classification feature information of the processed dynamic features is extracted through a recurrent neural network.

[0017] Preferably, the first classification feature information and the second classification feature information are concatenated and fused, and the category probability result after the fusion of the application shelf products is output in the trained classifier model includes:

[0018] Concatenate and fuse the corresponding vectors in the first classification feature information and the second classification feature information;

[0019] The fused classification feature information is output through the trained classifier model as the probability of each business category to obtain the category probability result.

[0020] Preferably, the fused classification feature information is output as the probability of each business category through the trained classifier model to obtain the category probability result, including:

[0021] Load a trained classifier model with output categories through the machine learning library;

[0022] The fused classification feature information is passed as input to the classifier model, and the prediction function of the classifier model is called to calculate the probability of each business category;

[0023] The classifier model outputs the probability of each business category into a two-dimensional array, each row of the two-dimensional array is a business category, and each column of the two-dimensional array is a predicted probability of a business category;

[0024] Generate category probability results according to the two-dimensional array.

[0025] Preferably, the matching of the deployment strategy corresponding to the application product in the deployment strategy template library according to the category probability result includes:

[0026] Constructing a deployment strategy matching each category in the classifier according to the classifier, and establishing a deployment strategy template library;

[0027] The service category with the highest probability in the category probability results is obtained, and the corresponding deployment strategy is automatically matched from the deployment strategy template library.

[0028] Preferably, the step of obtaining static features and dynamic features of application products extracted from the power dispatching edge cluster includes:

[0029] Obtain the static features of the application products after they are sent to the power dispatch edge cluster, including name, application type, version number, dependencies, and module composition;

[0030] The dynamic characteristics of the application product after running in the power dispatching edge cluster are obtained, including response time, CPU occupancy, memory usage and network traffic.

[0031] The present invention also provides an application product intelligent management system for an electric power dispatching edge cluster based on deep learning, which is applied to the aforementioned application product intelligent management method for an electric power dispatching edge cluster based on deep learning, including:

[0032] A feature data acquisition unit, used to acquire static features and dynamic features of application products extracted from the power dispatching edge cluster;

[0033] A classification feature extraction unit, used for extracting first classification feature information of the static feature and second classification feature information of the dynamic feature through a neural network;

[0034] A category probability output unit, used to concatenate and fuse the first classification feature information and the second classification feature information, and output the category probability result after the fusion of the application shelf products in the trained classifier model;

[0035] The strategy matching unit is used to match the deployment strategy corresponding to the application product in the deployment strategy template library according to the category probability result.

[0036] The present invention also provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the aforementioned deep learning-based intelligent management method for application products of the power dispatching edge cluster.

[0037] The present invention also provides a processor, which is used to run a program, wherein the program executes the aforementioned deep learning-based intelligent management method for application products of an electric power dispatching edge cluster when running.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The present invention provides an intelligent management method for application products of an electric power dispatching edge cluster based on deep learning by obtaining static features and dynamic features of application products extracted from the electric power dispatching edge cluster; extracting first classification feature information of the static features and second classification feature information of the dynamic features through a neural network; splicing and fusing the first classification feature information and the second classification feature information, and outputting the category probability result of the fused application products in the trained classifier model; matching the deployment strategy corresponding to the application products in the deployment strategy template library according to the category probability result. The present invention realizes intelligent management of application product deployment by automatically extracting and analyzing static and dynamic features of application products, using neural network models for classification and prediction, and finally generating reasonable deployment strategies, thereby improving the rationality of system functional module deployment and providing strong support for the efficient operation of edge clusters. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0041] Figure 1 This is a flow chart of an intelligent management method for application products of a power dispatching edge cluster based on deep learning of the present invention.

[0042] Figure 2 This is a schematic diagram of an intelligent management system for application products of a power dispatching edge cluster based on deep learning according to the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0045] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0046] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0047] See also Figure 1 The embodiment of the present application provides an intelligent management method for application products of a power dispatching edge cluster based on deep learning, including:

[0048] Step S01, obtaining static features and dynamic features of application products extracted from the power dispatching edge cluster;

[0049] After the power dispatching edge cluster receives the application product, it obtains the static features of the application product. The static features include text feature data such as name, application type, version number, dependencies and module composition. These feature data are usually related to static information such as metadata, configuration files, code structure, etc. of the application product, and do not involve actual runtime data.

[0050] After the application product runs in the power dispatching edge cluster, the dynamic characteristics of the application product after running in the power dispatching edge cluster are obtained, including digital data such as response time, CPU occupancy, memory usage and network traffic. These characteristic data reflect the dynamic information such as the behavior, performance, resource consumption, etc. of the application product during runtime.

[0051] Step S02, extracting first classification feature information of the static feature and second classification feature information of the dynamic feature through a neural network;

[0052] Specifically, the first classification feature information of the static feature extracted by using a neural network includes:

[0053] The data of the static features are subjected to preprocessing processes such as normalization, cropping, and denoising;

[0054] The processed static features are converted from high-dimensional sparse features of text to low-dimensional dense vectors of numbers through an embedding layer, and first classification feature information of the static features is extracted through a convolutional neural network;

[0055] Since static features are usually more intuitive and have low dimensions, they need to be reduced in dimension and encoded through an embedding layer. The digital form data of the static features after dimensionality reduction and encoding is input into the neural network for processing. At this time, the first classification feature information extracted from the neural network is vector feature data.

[0056] Specifically, the second classification feature information of the dynamic feature extracted by using a neural network includes:

[0057] Preprocessing the data of the dynamic feature to convert the dynamic feature into a vector of fixed length;

[0058] Dynamic features such as resource usage and application load are usually time series data or real-time data, which change over time and contain rich information. When processing this data to input into machine learning models such as neural networks, a series of preprocessing steps including standardization, normalization, and converting time series data into fixed-length vectors using sliding window technology are required to ensure data consistency and model effectiveness.

[0059] The second classification feature information of the processed dynamic features is extracted through a recurrent neural network.

[0060] At this time, the second classification feature information is vector feature data.

[0061] Step S03, concatenating and fusing the first classification feature information and the second classification feature information, and outputting the category probability result after the fusion of the application shelf products in the trained classifier model;

[0062] Preferably, the first classification feature information and the second classification feature information are concatenated and fused, and the category probability result after the fusion of the application shelf products is output in the trained classifier model includes:

[0063] Concatenate and fuse the corresponding vectors in the first classification feature information and the second classification feature information;

[0064] In the processing of the previous step, the first classification feature information and the second classification feature information have been converted into a vector form, and the two feature vectors are adjusted to have the same dimension, and then the two feature vectors are concatenated in the corresponding dimension.

[0065] The fused classification feature information is output as the probability of each business category through the trained classifier model. To obtain the category probability result, the trained classifier model needs to be loaded.

[0066] Specifically, the trained classifier model outputs the probability of each business category using the fused classification feature information, and the obtained category probability results include:

[0067] Load a trained classifier model with output categories through a machine learning library (such as scikit-learn, TensorFlow, PyTorch, etc.);

[0068] Among them, the classifier model is trained based on the data set, using a large amount of application product data with known classifications as the training set, so that the model can accurately classify new application products and can process the spliced ​​feature vectors. The spliced ​​feature vectors are input into the classifier model, and the model will output the probability of each business category. According to the probability results output by the model, it can be determined which business category the input application shelf product is most likely to belong to.

[0069] The fused classification feature information is passed as input to the classifier model, and the prediction function of the classifier model is called to calculate the probability of each business category;

[0070] The fused feature information can be a NumPy array, which is passed as input to the classifier model.

[0071] The classifier model outputs the probability of each business category into a two-dimensional array, each row of the two-dimensional array is a business category, and each column of the two-dimensional array is a predicted probability of a business category;

[0072] In this example, the business categories include standards such as application-based business type (such as Web services, big data processing, real-time analysis, etc.), resource requirements (such as CPU intensive, memory intensive, I / O intensive), and security level (such as low, medium, and high). For details, please refer to the probability results of each business category shown in Table 1.

[0073] Table 1 Two-dimensional array results of the probability of each business category

[0074]

[0075] Generate category probability results according to the two-dimensional array.

[0076] Step S04: Match the deployment strategy corresponding to the application product in the deployment strategy template library according to the category probability result.

[0077] Constructing a deployment strategy matching each category in the classifier according to the classifier, and establishing a deployment strategy template library;

[0078] According to the classification results of the neural network model, the most suitable deployment strategy template is automatically matched from the deployment strategy template library. However, since each product has its own uniqueness, the matching template may need to be customized to meet specific needs. The deployment strategy may include resource allocation schemes (such as the allocation ratio of CPU, memory, and storage), deployment location selection (such as which edge node to choose for deployment), load balancing strategy, security configuration, etc. The details are as follows:

[0079] Resource allocation plan: Dynamically adjust the allocation ratio of CPU, memory, and storage based on the resource demand forecast of the application product. For resource-sensitive applications, more resources may need to be allocated to ensure performance. Deployment location selection: Consider the geographical location, network conditions, resource availability and other factors of each node in the edge cluster, and select the most suitable node for deployment. For latency-sensitive applications, it may be necessary to select a node closer to the user to reduce network latency.

[0080] Load balancing strategy: Select an appropriate load balancing algorithm based on factors such as the number of concurrent requests and request patterns of the application product to balance the load of each node and improve the overall throughput of the system.

[0081] Security configuration: According to the security level and security requirements of the application product, configure the corresponding security policies, such as access control, data encryption, firewall rules, etc. For applications with high security levels, more stringent security measures may be required.

[0082] The service category with the highest probability in the category probability results is obtained, and the corresponding deployment strategy is automatically matched from the deployment strategy template library.

[0083] The customized deployment strategy is integrated into the automated deployment tool of the power dispatch edge cluster. The deployment tool pulls the largest business category application products, installs and configures the corresponding software and environment on the selected nodes, and then starts the application.

[0084] It should be noted that when selecting standardization or normalization methods for static and dynamic features, the distribution characteristics of the data and the requirements of the model should be considered. For example, if there are extreme outliers in the data set and these outliers have little impact on the model, then the standardization method can be selected. On the contrary, if there are no extreme outliers in the data set, or the outliers have a greater impact on the model, then the normalization method can be selected.

[0085] The present invention provides an intelligent management method for application products of an electric power dispatching edge cluster based on deep learning by obtaining static features and dynamic features of application products extracted from the electric power dispatching edge cluster; extracting first classification feature information of the static features and second classification feature information of the dynamic features through a neural network; splicing and fusing the first classification feature information and the second classification feature information, and outputting the category probability result of the fused application products in the trained classifier model; matching the deployment strategy corresponding to the application products in the deployment strategy template library according to the category probability result. The present invention realizes intelligent management of application product deployment by automatically extracting and analyzing static and dynamic features of application products, using neural network models for classification and prediction, and finally generating reasonable deployment strategies, thereby improving the rationality of system functional module deployment and providing strong support for the efficient operation of edge clusters.

[0086] See also Figure 2 The embodiment of the present application also provides an application product intelligent management system for an electric power dispatching edge cluster based on deep learning, which is applied to the aforementioned application product intelligent management method for an electric power dispatching edge cluster based on deep learning, including:

[0087] A feature data acquisition unit, used to acquire static features and dynamic features of application products extracted from the power dispatching edge cluster;

[0088] A classification feature extraction unit, used for extracting first classification feature information of the static feature and second classification feature information of the dynamic feature through a neural network;

[0089] A category probability output unit, used to concatenate and fuse the first classification feature information and the second classification feature information, and output the category probability result after the fusion of the application shelf products in the trained classifier model;

[0090] The strategy matching unit is used to match the deployment strategy corresponding to the application product in the deployment strategy template library according to the category probability result.

[0091] In this embodiment, the functional explanation corresponding to each unit is the same as the explanation of the intelligent management method of application products of a power dispatching edge cluster based on deep learning in the aforementioned embodiment, and the technical effect is also the same, so it will not be repeated here.

[0092] The present invention also provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the aforementioned deep learning-based intelligent management method for application products of the power dispatching edge cluster.

[0093] The technical effect of this embodiment is the same as the technical effect of an intelligent management method for application products of a power dispatching edge cluster based on deep learning in an embodiment, and will not be repeated here.

[0094] The present invention can be used in many general or special computer system environments or configurations, such as personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like.

[0095] The present invention also provides a processor, which is used to run a program, wherein the program executes the aforementioned deep learning-based intelligent management method for application products of an electric power dispatching edge cluster when running.

[0096] The technical effect of this embodiment is the same as the technical effect of the embodiment of a method for intelligent management of application products of an electric power dispatching edge cluster based on deep learning, and will not be repeated here.

[0097] The processor in this embodiment may be a central processing unit (CPU), a controller, a microcontroller, or other data processing chips.

[0098] Those of ordinary skill in the art will appreciate that the units of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0099] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0100] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0101] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, 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, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nlyMemory), random access memory (RAM, RandomAccessMemory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A method for intelligent management of application products of power dispatch edge cluster based on deep learning, characterized in that: include: Obtain static and dynamic features of application products extracted from the power dispatch edge cluster; Extracting first classification feature information of the static feature and second classification feature information of the dynamic feature through a neural network; The first classification feature information and the second classification feature information are concatenated and fused, and the category probability result after the fusion of the application shelf products is output in the trained classifier model; According to the category probability results, the deployment strategy corresponding to the application product is matched in the deployment strategy template library.

2. According to the method for intelligent management of application products of power dispatching edge cluster based on deep learning in claim 1, it is characterized in that: The first classification feature information of the static feature extracted by using a neural network includes: Preprocessing the data of the static features; The processed static features are converted from high-dimensional sparse features of text types to low-dimensional dense vectors of numbers through an embedding layer, and the first classification feature information of the static features is extracted through a convolutional neural network.

3. The method for intelligent management of application products of power dispatching edge cluster based on deep learning according to claim 2 is characterized in that: The second classification feature information of the dynamic feature extracted by using a neural network includes: Preprocessing the data of the dynamic feature to convert the dynamic feature into a vector of fixed length; The second classification feature information of the processed dynamic features is extracted through a recurrent neural network.

4. The method for intelligent management of application products of power dispatching edge cluster based on deep learning according to claim 3 is characterized in that: The first classification feature information and the second classification feature information are concatenated and fused, and the category probability result after the fusion of the application shelf products is output in the trained classifier model includes: Concatenate and fuse the corresponding vectors in the first classification feature information and the second classification feature information; The fused classification feature information is output through the trained classifier model as the probability of each business category to obtain the category probability result.

5. The method for intelligent management of application products of power dispatching edge cluster based on deep learning according to claim 4 is characterized in that: The trained classifier model outputs the probability of each business category by merging the classification feature information, and the obtained category probability results include: Load a trained classifier model with output categories through the machine learning library; The fused classification feature information is passed as input to the classifier model, and the prediction function of the classifier model is called to calculate the probability of each business category; The classifier model outputs the probability of each business category into a two-dimensional array, each row of the two-dimensional array is a business category, and each column of the two-dimensional array is a predicted probability of a business category; Generate category probability results according to the two-dimensional array.

6. The method for intelligent management of application products of power dispatch edge cluster based on deep learning according to claim 5 is characterized in that: The deployment strategy corresponding to the application product in the deployment strategy template library according to the category probability result includes: A deployment strategy matching each category in the classifier is constructed according to the classifier, and a deployment strategy template library is established; the business category with the highest probability in the category probability results is obtained, and the corresponding deployment strategy is automatically matched from the deployment strategy template library.

7. The method for intelligent management of application products of power dispatch edge cluster based on deep learning according to claim 1 is characterized in that: The acquisition of static features and dynamic features of application products extracted from the power dispatch edge cluster includes: Obtain the static features of the application products after they are sent to the power dispatch edge cluster, including name, application type, version number, dependencies, and module composition; The dynamic characteristics of the application product after running in the power dispatching edge cluster are obtained, including response time, CPU occupancy, memory usage and network traffic.

8. An intelligent management system for application products of power dispatch edge cluster based on deep learning, characterized in that: The method for intelligent management of application products of a power dispatching edge cluster based on deep learning applied to any one of claims 1 to 7 comprises: A feature data acquisition unit, used to acquire static features and dynamic features of application products extracted from the power dispatching edge cluster; a classification feature extraction unit, used to extract first classification feature information of the static features and second classification feature information of the dynamic features through a neural network; A category probability output unit, used to concatenate and fuse the first classification feature information and the second classification feature information, and output the category probability result after the fusion of the application shelf products in the trained classifier model; The strategy matching unit is used to match the deployment strategy corresponding to the application product in the deployment strategy template library according to the category probability result.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the application product intelligent management method of the power dispatching edge cluster based on deep learning as described in any one of claims 1 to 7.

10. A processor, characterized in that: The processor is used to run a program, wherein when the program is running, the method for intelligent management of application products of a power dispatching edge cluster based on deep learning as described in any one of claims 1 to 7 is executed.

Images