A big data processing method, system and storage medium for a cloud platform
Through the big data processing system of Xinchuang Cloud Platform, the user demand analysis module and software allocation module are used to build a user demand prediction model and generate the optimal software deployment solution, solving the problem of complexity of cloud management software deployment and achieving rapid and smooth The deployment process reduces the difficulty of work for technicians.
Patent Information
- Application Number
- CN202410187254.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-02-20
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-02-20
AI Technical Summary
The existing cloud management software has high requirements for technicians during the deployment and update process, and cannot achieve fast and smooth results, resulting in difficulties in enterprise data analysis and increasing the complexity of construction and deployment.
A big data processing system of Xinchuang Cloud Platform was designed, including user demand analysis module, software allocation module and user feedback information collection module. By building a user demand prediction model, the optimal software deployment plan is generated, and the application software is automatically installed to reduce the The difficulty of software selection and deployment.
By automatically generating the information of the software required by users and the optimal software deployment plan, the difficulty of software selection and deployment of large-scale data is effectively reduced, and the efficiency of cloud management software construction and deployment is improved.
Smart Images

Figure CN118012449B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of big data technology, and more specifically, to a big data processing method, system and storage medium for an innovative cloud platform. Background Art
[0002] Currently, common cloud platform management software is basically optimized and packaged based on the open source OpenStack architecture. Since cloud management software needs to uniformly manage cloud computing, storage, network and other resources, the required management functional components are very numerous, and the relationship between functional components is extremely complex. As a result, cloud management software has very high requirements for technical personnel in the deployment of application software and other links, and cannot achieve fast and smooth results. Technical personnel are required to analyze enterprise data, but enterprise data is huge, and technical personnel need to spend a lot of time analyzing enterprise data before they can understand the required software and software deployment, which makes the construction, deployment and update of cloud management software more difficult. Summary of the invention
[0003] The purpose of the present invention is to provide a big data processing method, system and storage medium for an innovative cloud platform in order to solve the above-mentioned problems.
[0004] The present invention provides a big data processing system for a cloud platform, comprising:
[0005] User demand analysis module, which is used to analyze user needs for software, including:
[0006] Build a user demand prediction model, obtain data of multiple users, and process the data of multiple users and input it into the user demand prediction model for training;
[0007] Obtain the data of the new user and create a software stack instance on the Xinchuang cloud platform based on the data of the new user;
[0008] After processing the data of new users, the data is input into the user demand prediction model to generate user prediction demand data;
[0009] Feedback the user's predicted demand data to the user terminal, obtain user feedback information, and determine whether to run the software stack instance to automatically install the application software based on the user feedback information;
[0010] The software allocation module is used to extract the type information of the software required by the user and generate the optimal software deployment plan, including:
[0011] Step 101, randomly generating a software code set, the software code set including a plurality of software codes;
[0012] Step 102, initializing the number of iterations to 0;
[0013] Step 103, calculating the fitness of the software code and recording the optimal software code;
[0014] Step 104: if the number of iterations is less than N, proceed to step 105; otherwise, directly output the optimal software code in step 103;
[0015] Step 105, generate a random number p and update the algorithm convergence factor And update the coefficient vector according to the formula and
[0016] Step 106, after updating the software code, return to step 103;
[0017] The user feedback information collection module is used to send a software running status feedback form to the user and obtain user feedback information.
[0018] As a further optimization scheme of the present invention, a user demand prediction model is constructed, data of multiple users are obtained, and the data of multiple users are processed and input into the user demand prediction model for training, specifically,
[0019] Obtain multiple user data and set the feature extraction range;
[0020] Different feature extraction methods are used for different feature types to extract corresponding feature data;
[0021] The extracted feature data is divided into a training set and a test set. The training set is used to train the user demand prediction model, and then the test set is used to evaluate the trained model.
[0022] When training the user demand prediction model, the log-likelihood function is used, and the expression is as follows:
[0023]
[0024] As a further optimization scheme of the present invention, the data of new users is processed and input into the user demand prediction model to generate user predicted demand data, specifically, setting the feature extraction range;
[0025] Different feature extraction methods are used for different feature types to extract corresponding feature data;
[0026] The extracted feature data is input into the user demand prediction model to generate user predicted demand data.
[0027] The expression of the user demand prediction model is:
[0028] h θ (x) = g(θ T X)
[0029] Among them, θ represents the model parameters, x represents the input feature vector, and g represents the sigmoid function;
[0030]
[0031] When h θ When (x) is greater than or equal to 0.5, the model predicts a positive class, which means that the user has a demand for the corresponding software. θ When (x) is less than 0.5, the model predicts a negative class, which means that the user has no demand for the corresponding software.
[0032] As a further optimization solution of the present invention, the type information of the software required by the user includes a software code, and the software code is composed of a plurality of unit software codes spliced in sequence, and the value of the i-th component of the i-th unit software code represents the software code of the i-th type of software allocated to the i-th user;
[0033] The type of software is accompanied by an ID in a fixed order, and the components in the unit software code are sorted according to this order;
[0034] The software code needs to meet the following constraints:
[0035] For the i-th unit software code, the value range of the i-th component value is the set of software codes of all i-th category software required by the i-th user.
[0036] As a further optimization solution of the present invention, in step 103, the method for calculating the fitness is specifically as follows:
[0037] Extracting information of corresponding software according to the software code represented by the software code component, simulating deployment on the cloud environment according to the information, and then obtaining cloud resources allocated to each software according to the result of the simulated deployment;
[0038] The formula for calculating fitness is:
[0039]
[0040] P i,j represents the allocation value of the jth resource of the i-th software deployed, Z i,j Indicates the maximum requirement value of the jth resource of the i-th software deployed.
[0041] As a further optimization scheme of the present invention, the algorithm convergence factor Contains the convergence factor a1 of the better subgroup and the convergence factor a2 of the worse subgroup, calculated as follows:
[0042]
[0043] a2=2·e -t / E
[0044] Coefficient vector and The update formula is:
[0045]
[0046] Where f(j) is the fitness value of the jth software code in the software code set, and f(gbest) is the fitness value of the best software code in the software code set.
[0047] As a further optimization solution of the present invention, the method for updating the software code in step 106 includes:
[0048] If p is less than 0.5, determine whether |A| is less than 1. If |A| is less than 1, update the software code using the first formula;
[0049] The first formula is as follows:
[0050]
[0051] In the formula, Indicates the software code with the best objective function value in the current software code set, Indicates the enclosing step length of the current software code;
[0052] If p is less than 0.5 and |A| is not less than 1, the software code is updated using the second formula;
[0053] The second formula is as follows:
[0054]
[0055] In the formula, Indicates a software code randomly selected from the current software code set. is the current software code, represents the distance vector between the current software code and the randomly selected software code,
[0056] If p is greater than or equal to 0.5, the software code is updated using the third formula;
[0057] The third formula is as follows:
[0058]
[0059] In the formula, Represents the distance vector between the current software code and the optimal software code, the constant coefficient b is 1, and l is a random number in the interval [-1,1].
[0060] As a further optimization scheme of the present invention, the calculation formula of ω is as follows:
[0061]
[0062] In the formula, ω max is the maximum initial value of the inertia weight, ω max =0.9,ω min is the minimum initial value of inertia weight, ω min =0.4.
[0063] A big data processing method for a cloud computing platform, comprising:
[0064] Step 201, constructing a user demand prediction model, obtaining data of multiple users, and processing the data of multiple users and inputting the data into the user demand prediction model for training;
[0065] Step 202, obtaining data of the new user, and creating a software stack instance on the Xinchuang cloud platform based on the data of the new user;
[0066] Step 203, after processing the data of the new user, input it into the user demand prediction model and generate user predicted demand data;
[0067] Step 204, feeding back the user predicted demand data to the user terminal, obtaining user feedback information, and determining whether to run the software stack instance to automatically install the application software based on the user feedback information;
[0068] Step 205, randomly generating a software code set, the software code set including a plurality of software codes;
[0069] Step 206, initializing the number of iterations to 0;
[0070] Step 207, calculating the software code fitness and recording the optimal software code;
[0071] Step 208, if the number of iterations is less than N, proceed to step 209, otherwise directly output the optimal software code in step 207;
[0072] Step 209, generate a random number p, update the algorithm convergence factor, and update the coefficient vector according to the formula and
[0073] Step 210, after updating the software code, return to step 207;
[0074] Step 211, sending a software running status feedback form to the user and obtaining user feedback information.
[0075] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned big data processing method for an innovative cloud platform.
[0076] The beneficial effect of the present invention is that the present invention can automatically generate information about the software required by the user and the optimal software deployment plan after analyzing and processing the data of the enterprise user, which can effectively reduce the difficulty of the user in selecting and deploying software for large quantities of data. BRIEF DESCRIPTION OF THE DRAWINGS
[0077] Figure 1 It is a system module diagram of the present invention;
[0078] Figure 2 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION
[0079] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only intended to enable those skilled in the art to better understand and implement the subject matter described herein. In addition, the features described relative to some examples may also be combined in other examples.
[0080] like Figure 1 As shown, a big data processing system of a Xinchuang cloud platform includes:
[0081] User demand analysis module, which is used to analyze user needs for software, including:
[0082] Build a user demand prediction model, obtain data of multiple users, and process the data of multiple users and input it into the user demand prediction model for training;
[0083] Obtain the data of the new user and create a software stack instance on the Xinchuang cloud platform based on the data of the new user;
[0084] After processing the data of new users, the data is input into the user demand prediction model to generate user prediction demand data;
[0085] Feedback the user's predicted demand data to the user terminal, obtain user feedback information, and determine whether to run the software stack instance to automatically install the application software based on the user feedback information.
[0086] Based on the new user's data, a software stack instance is created on the Xinchuang Cloud platform. Specifically,
[0087] Log in to the Xinchuang Cloud Platform: Use your account and password to log in to the Xinchuang Cloud Platform console;
[0088] Select an application stack: Find the corresponding menu or link in the console and select "Application Stack" or a similar option;
[0089] Create an instance: On the interface of the application software stack, there is usually a button or link called "Create instance" or "New instance". Click it to enter the page for creating an instance.
[0090] Configure instance parameters: On the instance creation page, automatically enter the basic parameters of the instance, including instance name, region, availability zone, instance specifications (CPU, memory, storage capacity, etc.), operating system type, and other configuration options;
[0091] Network and security configuration: Configure the instance's network settings, security group rules, and access permissions based on the user's historical data to ensure that the instance can safely provide external services and communicate with other components;
[0092] Deploy application software: According to the requirements of the application software stack, select the specific application software and components to be deployed, such as database, application server, load balancer, etc., and configure them accordingly;
[0093] Start the instance: After confirming that the configuration is correct, click the "Create" or "Confirm" button to start the instance creation process and create the corresponding software stack instance on the Xinchuang Cloud Platform;
[0094] Monitoring and management: After the instance is successfully created, it is monitored and managed to ensure that the instance is running normally.
[0095] The following is the program code for creating a software stack instance using the API provided by the platform: import
[0096]
[0097]
[0098] Build a user demand prediction model, obtain data from multiple users, and process the data of multiple users before inputting it into the user demand prediction model for training. Specifically,
[0099] Obtain multiple user data and set the feature extraction scope, which includes but is not limited to the user's personal information, transaction history, product usage, and online behavior;
[0100] Different feature extraction methods are used for different feature types to extract corresponding feature data, for example:
[0101] Statistical feature extraction: For user personal information and some simple behavioral data (such as the number of purchases, the number of comments, etc.), feature extraction can be performed through statistical methods (such as mean, standard deviation, maximum, minimum, etc.);
[0102] Text feature extraction: For some text data (such as search keywords, comment content, etc.), natural language processing technology can be used for feature extraction, including bag-of-words model, TF-IDF model, topic model, etc.
[0103] Image feature extraction: For some image data (such as user avatars, uploaded pictures, etc.), computer vision technology can be used for feature extraction, including color histograms, local binary patterns, Haar features, deep learning features, etc.;
[0104] Time series feature extraction: For some time series data (such as user behavior time, access frequency, etc.), time series analysis techniques can be used for feature extraction, including autoregressive models, sliding window statistics, etc.
[0105] The extracted feature data is divided into a training set and a test set. The training set is used to train the user demand prediction model, and then the test set is used to evaluate the trained model.
[0106] After processing the data of new users, input it into the user demand prediction model and generate user prediction demand data, specifically,
[0107] Obtain new user data and set the feature extraction scope, which includes but is not limited to the user's personal information, transaction history, product usage, and online behavior;
[0108] Different feature extraction methods are used for different feature types to extract corresponding feature data;
[0109] The extracted feature data is input into the user demand prediction model to generate user predicted demand data.
[0110] The expression of the user demand prediction model is:
[0111] h θ (x) = g(θ T X)
[0112] Among them, θ represents the model parameters, x represents the input feature vector, and g represents the sigmoid function;
[0113]
[0114] When h θ When (x) is greater than or equal to 0.5, the model predicts a positive class, which means that the user has a demand for the corresponding software. θ When (x) is less than 0.5, the model predicts a negative class, which means that the user has no demand for the corresponding software.
[0115] When training the user demand prediction model, the log-likelihood function is used, and the expression is as follows:
[0116]
[0117] For example, when corresponding feature data appears in the data of a new user, the rate of change of the feature data over time series is calculated based on the number of times the feature data appears in the interval time period, and the user demand prediction model is input. If the output value is greater than 0.5, it means that the new user also needs corresponding software to manage or process certain specific data. Otherwise, the corresponding software is not needed, but it is still necessary to feedback to the client and obtain the customer's final actual needs. This can effectively provide customers with data changes that they have not noticed and recommend corresponding application software to reduce the customer's workload and work difficulty.
[0118] Based on the user feedback information, it is determined whether to run the software stack instance to automatically install the application software. Specifically, if the user feedback information indicates that the corresponding application software needs to be installed, the software stack instance is run to install the software.
[0119] Run the software stack instance to install the software, specifically,
[0120] Launch a software stack instance;
[0121] Create the resources and environment required for running application software in the cloud environment;
[0122] Manage the configuration required for application software installation;
[0123] Automatically install application software;
[0124] Run monitoring programs;
[0125] Configure the running parameters of the application software.
[0126] The software allocation module extracts the type information of the software required by the user. A user may require multiple software, and the type information of each software required by the user needs to be extracted;
[0127] The software code is composed of a plurality of unit software codes spliced in sequence, and the value of the i-th component of the i-th unit software code represents the software code of the i-th category of software allocated to the i-th user;
[0128] The type of software is accompanied by an ID in a fixed order, and the components in the unit software code are sorted according to this order;
[0129] The software code needs to meet the following constraints:
[0130] For the i-th unit software code, the value range of the i-th component value is the set of software codes of all i-th category software required by the i-th user;
[0131] In one embodiment of the present invention, the software required by the user is manually selected by the user according to his / her own needs.
[0132] In one embodiment of the present invention, the software required by the user is the result of analysis by the user demand analysis module.
[0133] Step 101, randomly generating a software code set, the software code set including a plurality of software codes;
[0134] Step 102, initializing the number of iterations to 0;
[0135] Step 103, calculate the software code fitness and record the optimal software code. The fitness calculation method is specifically as follows:
[0136] Extracting information of corresponding software according to the software code represented by the software code component, simulating deployment on the cloud environment according to the information, and then obtaining cloud resources allocated to each software according to the result of the simulated deployment;
[0137] The formula for calculating fitness is:
[0138]
[0139] P i,j represents the allocation value of the jth resource of the i-th software deployed, Z i,j Indicates the maximum demand value of the jth resource of the i-th software deployed;
[0140] Resources include bandwidth, storage, memory, threads, etc.
[0141] Step 104: if the number of iterations is less than N, proceed to step 105; otherwise, directly output the optimal software code in step 103;
[0142] Step 105, generate a random number p and update the algorithm convergence factor And update the coefficient vector according to the formula and
[0143] Algorithm convergence factor Contains the convergence factor a1 of the better subgroup and the convergence factor a2 of the worse subgroup, calculated as follows:
[0144]
[0145] a2=2·e -t / E
[0146] Coefficient vector and The update formula is:
[0147]
[0148] Where, f(j) is the fitness value of the jth software code in the software code set, and f(gbest) is the fitness value of the best software code in the software code set;
[0149] Step 106, after updating the software code, return to step 103; the method for updating the software code includes:
[0150] If p is less than 0.5, determine whether |A| is less than 1. If |A| is less than 1, update the software code using the first formula;
[0151] The first formula is as follows:
[0152]
[0153] In the formula, Indicates the software code with the best objective function value in the current software code set, Indicates the enclosing step length of the current software code;
[0154] If p is less than 0.5 and |A| is not less than 1, the software code is updated using the second formula;
[0155] The second formula is as follows:
[0156]
[0157] In the formula, Indicates a software code randomly selected from the current software code set. is the current software code, represents the distance vector between the current software code and the randomly selected software code,
[0158] If p is greater than or equal to 0.5, the software code is updated using the third formula;
[0159] The third formula is as follows:
[0160]
[0161] In the formula, Represents the distance vector between the current software code and the optimal software code, the constant coefficient b is 1, and l is a random number in the interval [-1,1].
[0162] The calculation formula of the aforementioned ω is as follows:
[0163]
[0164] In the formula, ω max is the maximum initial value of the inertia weight, ω max =0.9,ωmin is the minimum initial value of inertia weight, ω min =0.4.
[0165] In one embodiment of the present invention, the method for randomly generating a software code set includes:
[0166] Set the size of the software code set to N, the dimension of the software code in the software code set to d, the maximum number of chaotic iteration steps to k, and the control parameter α = 0.489;
[0167] Randomly generate a d-dimensional software code;
[0168] The initial d-dimensional software code is iterated 400 times using Kent mapping to obtain the initial software code of the software code set;
[0169] Use Kent mapping to iterate each dimensional element of the initial software code of the software code set, and finally generate an N*d matrix;
[0170] The chaotic variables are converted into design variables in the algorithm through inverse mapping, and the chaotic variables in the matrix are defined as x ij The upper and lower limits of the design variables in the algorithm are lb and ub respectively, and the corresponding actual variable is x ij =lb+(ub-lb)·x ij .
[0171] The user feedback information collection module is used to send software running status feedback forms to users and obtain real-time user feedback. It is mainly used to collect problems or directions for improvement in actual user use, so as to provide improvement parameters for subsequent software recommendations and deployment.
[0172] like Figure 2 As shown, the present invention also provides a big data processing method for a Xinchuang cloud platform, comprising:
[0173] Step 201, constructing a user demand prediction model, obtaining data of multiple users, and processing the data of multiple users and inputting the data into the user demand prediction model for training;
[0174] Step 202, obtaining data of the new user, and creating a software stack instance on the Xinchuang cloud platform based on the data of the new user;
[0175] Step 203, after processing the data of the new user, input it into the user demand prediction model and generate user predicted demand data;
[0176] Step 204, feeding back the user predicted demand data to the user terminal, obtaining user feedback information, and determining whether to run the software stack instance to automatically install the application software based on the user feedback information;
[0177] Step 205, randomly generating a software code set, the software code set including a plurality of software codes;
[0178] Step 206, initializing the number of iterations to 0;
[0179] Step 207, calculating the software code fitness and recording the optimal software code;
[0180] Step 208, if the number of iterations is less than N, proceed to step 209, otherwise directly output the optimal software code in step 207;
[0181] Step 209, generate a random number p, update the algorithm convergence factor, and update the coefficient vector according to the formula and
[0182] Step 210, after updating the software code, return to step 207;
[0183] Step 211, sending a software running status feedback form to the user and obtaining user feedback information.
[0184] The present invention also provides a computer-readable storage medium on which a computer program is stored, which, when executed by a processor, implements the big data processing method of the Xinchuang cloud platform as described above.
[0185] The present embodiment is described above, but the present embodiment is not limited to the above-mentioned specific implementation manner. The above-mentioned specific implementation manner is merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, which are all within the protection of the present embodiment.
Claims
1. A big data processing system for a cloud platform, characterized in that: include: User demand analysis module, which is used to analyze user needs for software, including: Build a user demand prediction model, obtain data of multiple users, and process the data of multiple users and input it into the user demand prediction model for training; Obtain the data of the new user and create a software stack instance on the Xinchuang cloud platform based on the data of the new user; After processing the data of new users, input it into the user demand prediction model and generate user predicted demand data, and set the feature extraction range; Different feature extraction methods are used for different feature types to extract corresponding feature data; Inputting the extracted feature data into a user demand prediction model and generating user predicted demand data; The expression of the user demand prediction model is: ; in, represents the model parameters, x represents the input feature vector, and g represents the sigmoid function; ; when When it is greater than or equal to 0.5, the user demand prediction model predicts a positive class, which means that the user has a demand for the corresponding software. When it is less than 0.5, the user demand prediction model predicts a negative class, which means that the user has no demand for the corresponding software; Feedback the user's predicted demand data to the user terminal, obtain user feedback information, and determine whether to run the software stack instance to automatically install the application software based on the user feedback information; The software allocation module is used to extract the type information of the software required by the user and generate the optimal software deployment plan, including: Step 101, randomly generating a software code set, the software code set including a plurality of software codes; Step 102, initializing the number of iterations to 0; Step 103, calculating the fitness of the software code and recording the optimal software code; Step 104: if the number of iterations is less than N, proceed to step 105; otherwise, directly output the optimal software code in step 103; Step 105, generate a random number p and update the algorithm convergence factor , and update the coefficient vector according to the formula and , algorithm convergence factor Convergence factor containing better subgroups and the convergence factor of the poor subgroup , the calculation formula is as follows: ; ; Coefficient vector and The update formula is: ; ; ; In the formula, is the fitness value of the jth software code in the software code set, is the fitness value of the optimal software code in the software code set; Step 106, after updating the software code, return to step 103; The user feedback information collection module is used to send a software running status feedback form to the user and obtain user feedback information.
2. According to the big data processing system of the Xinchuang cloud platform according to claim 1, it is characterized in that: Build a user demand prediction model, obtain data from multiple users, and process the data of multiple users before inputting it into the user demand prediction model for training. Specifically, Obtain multiple user data and set the feature extraction range; Different feature extraction methods are used for different feature types to extract corresponding feature data; The extracted feature data is divided into a training set and a test set. The training set is used to train the user demand prediction model, and then the test set is used to evaluate the trained user demand prediction model. When training the user demand prediction model, the log-likelihood function is used, and the expression is as follows: ; 。 3. According to the big data processing system of the Xinchuang cloud platform according to claim 1, it is characterized in that: The type information of the software required by the user includes a software code, which is composed of a plurality of unit software codes spliced in sequence, and the value of the i-th component of the i-th unit software code represents the software code of the i-th type of software allocated to the i-th user; The type of software is accompanied by an ID in a fixed order, and the components in the unit software code are sorted according to this order; The software code needs to meet the following constraints: For the i-th unit software code, the value range of the i-th component value is the set of software codes of all i-th category software required by the i-th user.
4. According to claim 1, a big data processing system for a trusted cloud platform is characterized in that: In step 103, the method for calculating the fitness is specifically as follows: Extracting information of corresponding software according to the software code represented by the software code component, simulating deployment on the cloud environment according to the information, and then obtaining cloud resources allocated to each software according to the result of the simulated deployment; The formula for calculating fitness is: ; represents the allocation value of the jth resource of the i-th software deployed, Indicates the maximum requirement value of the jth resource of the i-th software deployed.
5. According to the big data processing system of the Xinchuang cloud platform according to claim 1, it is characterized in that: The method for updating the software code in step 106 includes: If p is less than 0.5, Is it less than 1? If it is less than 1, the software code is updated by the first formula; The first formula is as follows: ; In the formula, Indicates the software code with the best objective function value in the current software code set, Indicates the enclosing step length of the current software code; If p is less than 0.5, If it is not less than 1, the software code is updated by the second formula; The second formula is as follows: ; In the formula, Indicates a software code randomly selected from the current software code set. is the current software code, represents the distance vector between the current software code and the randomly selected software code, If p is greater than or equal to 0.5, the software code is updated using the third formula; The third formula is as follows: ; In the formula, Represents the distance vector between the current software code and the optimal software code, the constant coefficient b is 1, and l is a random number in the interval [-1,1].
6. According to claim 5, a big data processing system for a trusted cloud platform is characterized in that: Said The calculation formula is as follows: ; In the formula, is the maximum initial value of the inertia weight, , is the minimum initial value of inertia weight, .
7. A big data processing method for a Xinchuang cloud platform, characterized in that: include: Step 201, constructing a user demand prediction model, obtaining data of multiple users, and processing the data of multiple users and inputting the data into the user demand prediction model for training; Step 202, obtaining data of the new user, and creating a software stack instance on the Xinchuang cloud platform based on the data of the new user; Step 203, after processing the data of the new user, input it into the user demand prediction model and generate user predicted demand data, and set the feature extraction range; Different feature extraction methods are used for different feature types to extract corresponding feature data; Inputting the extracted feature data into a user demand prediction model and generating user predicted demand data; The expression of the user demand prediction model is: ; in, represents the model parameters, x represents the input feature vector, and g represents the sigmoid function; ; when When it is greater than or equal to 0.5, the user demand prediction model predicts a positive class, which means that the user has a demand for the corresponding software. When it is less than 0.5, the user demand prediction model predicts a negative class, which means that the user has no demand for the corresponding software; Step 204, feeding back the user predicted demand data to the user terminal, obtaining user feedback information, and determining whether to run the software stack instance to automatically install the application software based on the user feedback information; Step 205, randomly generating a software code set, the software code set including a plurality of software codes; Step 206, initializing the number of iterations to 0; Step 207, calculating the software code fitness and recording the optimal software code; Step 208, if the number of iterations is less than N, proceed to step 209, otherwise directly output the optimal software code in step 207; Step 209, generate a random number p and update the algorithm convergence factor , and update the coefficient vector according to the formula and , algorithm convergence factor Convergence factor containing better subgroups and the convergence factor of the poor subgroup , the calculation formula is as follows: ; ; Coefficient vector and The update formula is: ; ; ; In the formula, is the fitness value of the jth software code in the software code set, is the fitness value of the optimal software code in the software code set; Step 210, after updating the software code, return to step 207; Step 211, sending a software running status feedback form to the user and obtaining user feedback information.
8. A computer-readable storage medium, characterized in that: A computer program is stored thereon, which, when executed by a processor, implements the big data processing method of the Xinchuang cloud platform as described in claim 7.
