CN116662116A

CN116662116A - Cloud platform-based mode training preparation one-key starting method

Info

Publication number: CN116662116A
Application number: CN202310625311.6A
Authority: CN
Inventors: 刘顺华; 徐权周; 宋斌; 李慧卓; 张博; 龙成亮; 逯璐; 樊志成; 刘金晶
Original assignee: Insititute Of Nbc Defence
Current assignee: Insititute Of Nbc Defence
Priority date: 2023-05-30
Filing date: 2023-05-30
Publication date: 2023-08-29

Abstract

The application discloses a cloud platform-based method for preparing one-key starting by using a module training method. The method specifically comprises the following steps: the cloud and end resources of the training center are abstracted uniformly; at a system access layer, through unified login, the system can access all software of a model training through one account, a computer is started by one key, and the states of a server and a client PC are monitored; after the hardware equipment is ready to be completed, the training system is started by one key, and the configured client software in the arrangement scheme is started by one key through a monitoring center to monitor the states of the client and the server software; monitoring the training process, quickly recovering the server after the background has problems, quickly replacing equipment when the client PC has problems, and sending down the soft parts again or starting a browser access system; training is resumed. The cloud resource virtual machine abstract method and device provided by the application are used for carrying out unified abstraction on resources of the cloud and the terminal, providing a unified resource view angle, and providing a more visual use view angle while enjoying the advantage of cloud resource virtual machine.

Description

Cloud platform-based mode training preparation one-key starting method

Technical Field

The application relates to the technical field of computers, in particular to a cloud platform-based method for preparing one-key starting by using a module training method.

Background

With the development of the modeling and training technology, the software composition of the system is evolved from original single-node single software or hardware training to the participation of a simulator composed of multiple nodes and multiple software and hardware; the use environment of the system is evolved from the original single operating system environment to a multi-operating system environment, including embedded operating system, windows operating system, linux operating system and other environments; the operation and maintenance scale of the system is calculated and formed from a plurality of classrooms, and is evolved into a multi-classroom, multi-server and multi-PC. With the improvement of the integration scale and complexity of the system, the traditional system operation and maintenance management method enables operation and maintenance personnel of the model training system to sink into a large number of trivial operation and maintenance guarantee works of the model training system, and a large amount of time is required from starting the system to finishing training preparation, so that the service efficiency of the model training system is seriously affected.

The existing various systems also perform some beneficial exploration under the non-cloud mode, such as a developed software package management function, so that the management of the C/S client software can be conveniently performed, the difficulty of software deployment and updating is reduced to a certain extent, but with the increase of software forms, such as server software, B/S client software, cloud desktop software and virtual machine software, the traditional C/S client management software cannot cope with the software, and a corresponding solution is provided.

The general use flow of the existing simulation training system is as follows:

(1) Starting training all client PCs;

(2) Configuring the seat configuration of each computing software according to the requirement and starting the software;

(3) Testing whether each seat software works normally;

(4) Providing the trainees with system software for training;

(5) The occurrence of the problem is reported by the students;

(6) And (3) operating and maintaining personnel solve the problem in the problem machine, if the computer has a problem and the client software is C/S software, the model training client software needs to be reinstalled after the computer is replaced.

The use flow of the existing system mainly has the following problems:

(1) The preparation time is too long, or the training time is shortened, or the manpower input is increased;

(2) The system state is opaque, and the problem is not reflected timely;

(3) The recovery time of the system is too long when a problem occurs;

(4) The software problems that can be handled are relatively single, and only some of the operating and maintenance problems of the C/S software can be handled.

By adopting the technical scheme of the application, the solution scheme of the existing scheme is as follows:

(1) The preparation time is too long, and the quick system preparation can be realized by starting up by one key and starting up the application by one key;

(2) The system state is opaque, the problem reaction is not timely, the system state is monitored in real time through unified monitoring of the cloud and the terminal, and the system problem is timely positioned;

(3) The system recovery time is too long, the cloud uses the high availability of the service to realize the rapid scheduling and recovery of the service, the client B/S application can reload the system through simple refreshing, the client B/S application restarts the application through a cloud assistant, and the application recovery can be realized by using the application deployment function of the cloud assistant after the computer is replaced for the node with the problem of the computer.

After solving the problems existing in the prior art, the application has the following technical effects:

(1) The computing power of the terminal can be brought in, the problem of insufficient computing power of the GPU of the cloud computing service terminal can be complemented by a C/S software and virtual machine software distribution technology, and the existing computing power is multiplexed to the greatest extent;

(2) Multiplexing the existing arrangement scheme, and being applicable to different training scenes;

(3) The cloud resource virtual machine has the advantages that resources of the cloud and the terminal are abstracted uniformly, a uniform resource view angle is provided, and a more visual use view angle is provided while enjoying the advantages of the cloud resource virtual machine.

Disclosure of Invention

In view of the above problems, the present application provides a method for preparing a one-key start for a model training of a cloud platform, comprising the following steps:

(1) Training to carry out resource arrangement or multiplexing the existing resource arrangement scheme;

(2) Starting the computer by one key, and monitoring the states of the server and the client PC;

(3) One-key starting application, monitoring the software states of a client and a server;

(4) Monitoring the training process, quickly recovering the server after the background has problems, quickly replacing equipment when the client PC has problems, and sending down the soft parts again or starting a browser access system;

(5) Training is resumed.

The step (1) specifically comprises the following steps:

integrating various devices in a resource abstraction layer to form the resource abstraction layer of the device, and providing a state monitoring function for one-key starting to know the system state after starting in real time;

in the application abstraction layer, application layer abstraction is needed, and client side B/S software, C/S software, cloud desktop application and virtual machine application are unified, and the four application forms can be combined with the capabilities of cloud and end to provide a deployment running environment for covering and training all types of software.

The cloud assistant collects system information of the edge PC, and in the training preparation stage, the guiding and controlling software can fetch a software list to deploy and arrange seats to form an application deployment scheme.

The cloud assistant has the following specific functions:

1) Image downloading: downloading a required mirror image from the cloud platform according to service requirements;

2) And (3) data downloading: according to the service requirement, downloading the required data from the cloud platform, comparing the local data with the server data according to the requirement, avoiding repeated downloading, and reducing the network pressure;

3) Virtual machine image downloading and starting: downloading the virtual machine image according to the service requirement, and starting the virtual machine image locally;

4) Cloud desktop starting and connecting: when the cloud desktop application is entered through the comprehensive portal, a local cloud assistant where the browser is located applies for corresponding cloud desktop resources, and automatically starts a cloud desktop client to connect with a remote cloud desktop;

5) The data is returned to the cloud platform: and in the training and withdrawing stage, the data are returned to the cloud platform for storage according to the service requirements.

The step (1) further specifically includes:

in order to cooperate with realizing the one-key preparation function on the client PC, a cloud assistant tool needs to be developed, and the cloud assistant supports the functions of mirror image downloading, environment data downloading, cloud desktop resource application, mirror image downloading and starting, and provides support for unified application management.

Training resource planning can be performed through a space view, and resources such as sites, seats, equipment, software and the like for training are planned.

And the one-key starting computer is used for starting all services and client computers which are depended on by the system according to the training scheme arranged in the training preparation stage, and the starting states of the trained computer and the simulator can be checked through the monitoring center after the system is started.

The application opening is unified by the following means:

(1) The Web application is directly opened through a browser;

(2) The cloud desktop application clicks on the browser, simultaneously triggers a background to prepare cloud desktop resources, simultaneously informs a user desktop of the current cloud desktop preparation state through a cloud assistant, and automatically logs in the prepared cloud desktop through carrying user login information by the cloud assistant after the cloud desktop preparation is completed;

(3) The physical machine application clicks on the browser and triggers a user desktop cloud assistant at the same time, if the local software does not exist, the software pulling is carried out, the software pulling state is displayed, if the local software exists, a local cache is used, and software starting is carried out after the software preparation is completed;

(4) And the physical virtual desktop application clicks on the browser and triggers a user desktop cloud assistant, if the local virtual machine mirror image does not exist, mirror image pulling is performed, the mirror image pulling state is displayed, if the local virtual machine mirror image exists, a local cache is used, and after the virtual machine mirror image preparation is completed, the local KVM is used for starting the mirror image.

After the training is finished, a one-key withdrawal function is provided, from the business, a group of trainees inform each piece of trained software of finishing the training through the guiding control system, each piece of trained software performs corresponding resource cleaning and data recovery, after the business is cleaned, the software closing, resource returning and computer closing can be completed through one key of the model training operation and maintenance system, after a one-key withdrawal instruction is sent, the withdrawal condition can be checked through the monitoring system, and the completion of the training withdrawal work after no abnormality is ensured.

Drawings

FIG. 1 shows a system hierarchy architecture schematic;

FIG. 2 illustrates an application orchestration and deployment schematic;

FIG. 3 shows a flow chart for a model training preparation one-key start.

Detailed Description

The application provides a training preparation one-key starting method based on a cloud platform, which aims at the characteristics of a training center cloud and end equipment, performs unified abstraction on the training center cloud and end resources, provides a resource arrangement function in a training subject and training plan view angle, and finally provides training B/S and C/S application software for group training and trained personnel, and presents complex concepts of the cloud and the end in an easy-to-understand use view angle.

At the resource abstraction layer, various devices accessed by module training, including devices such as a PC, a camera, a cabinet moving ring, a radio station, a simulator, an access control device, an audio and video device and the like, are integrated by adopting the gateway technology of the Internet of things, so that the resource abstraction layer of the device is formed, a state monitoring function can be provided for one-key starting, and the system state after starting is known in real time.

At an application abstraction layer, no matter a Web service and a cloud desktop running on computing resource resources on a cloud or a B/S client, C/S software and a virtual machine application running on computing resource on a terminal are required to be abstracted at the application layer, the B/S software, the C/S software, the cloud desktop application and the virtual machine application of the client are unified, and the four application forms can combine the capabilities of the cloud and the terminal to provide a deployment running environment covering and training all types of software.

In a system access layer, through unified login, the system can access all software of a model training through one account, the access authority problem of various arranged applications is solved, training subjects and training plan time are provided, a resource arrangement function is provided for each training plan, and convenient training operations such as resource planning, client software deployment, one-key preparation and the like can be performed according to the training requirements of subjects.

(1) The container service application, the system background service, the high-availability container service running in a cluster mode, most of simulation, guide and control and cloud platform services are developed and deployed in a container mode and exist in a Docker container mode.

(2) The cloud desktop application is used for lightweight training desktops and exists in a mode of virtual machine VM image.

(3) The local PC virtual machine application is used for applications with certain GPU rendering requirements such as a desktop training system, a subsystem and the like, solves the problem of environment isolation of single PC deployment of multiple applications, and exists in a virtual machine mode.

1. System architecture

As shown in fig. 1, the system architecture is mainly divided into four layers:

1. the device access layer is used for accessing various devices for supporting training, is equivalent to a perception layer of an architecture of the Internet of things, can access various devices, and has expandability;

2. the resource management layer is used for uniformly managing the accessed resources and providing resource support for the service module;

3. the resource monitoring layer monitors the accessed resources and provides a monitoring view for users;

4. the system service layer is used for carrying out service modeling on teaching related services and constructing a service module for supporting actual training;

5. and the system access layer provides access to portal capacity to the outside, manages system login through a unified authentication system and expands and accesses applications of various factories through desktop applications.

Device access stratum

Access door control, cameras, cabinet moving rings, sound equipment, computers, radio stations and simulator equipment are provided.

a. The access control equipment is accessed, the access control is uniformly managed through the large-scale platform, and the comprehensive portal is accessed through the large-scale platform.

b. The camera equipment is accessed, the security cameras are uniformly managed through the Dahua platform, and the comprehensive portal is accessed through the sea health platform.

c. And the cabinet moving ring equipment is accessed, and the comprehensive portal accesses alarm events such as air-conditioning states and humidity abnormal events through a WebService interface provided by the cabinet moving ring.

d. The sound equipment is connected, the sound serial port equipment is connected through the serial port to network port arrangement, and the comprehensive portal background service controls sound through a serial port communication data protocol.

e. The computer equipment is accessed, the online state and the application starting capability of the computer are accessed through a cloud assistant, and the desktop screen of the computer is accessed through a vnc protocol based on guaamaole.

f. The simulator equipment is accessed, accessed through the standard DDS protocol issued by the total director, sent by each classroom simulator,

g. the radio equipment is accessed, and the data forwarding is performed through a standard DDS protocol after the access through the private UDP protocol provided by a radio manufacturer.

(II) resource management layer

Providing a venue, a seat, a computer, software, and camera resource management.

a. The site resource is used for carrying out site association during training planning through a visual site visual input function.

b. The seat resources are visually input into the seats, and the fixed seats can be associated with simulators and computers.

c. Computer resources are visually recorded into a computer, and the computer is associated with an actual computer through IP and the computer is associated with a seat.

d. The software resource can be recorded into four types of software, namely local software, web software, virtual machine software and cloud desktop software through an application center.

e. The camera resources are input in a visual mode, and implementation and historical camera video streams can be called through equipment numbers, IP and bloom platform association.

(III) resource monitoring layer

a. The machine room resource monitoring can be connected with the machine cabinet movable ring monitoring through the monitoring alarm function, and the machine room abnormal event is monitored in real time.

b. The cloud platform resource monitoring can synchronously alarm the cloud platform alarm event on the portal by accessing the operation platform and the alarm interface thereof.

c. The classroom security monitoring can be realized by accessing the monitoring capability of the bloom platform, so that the video stream of each security camera can be called in real time, and the video stream can be displayed on a screen.

d. And monitoring desktop services, namely pulling a desktop video stream through a desktop monitoring gateway by using computer client monitoring software started by a cloud assistant, and pushing the desktop video stream to a portal for display.

e. The radio station monitoring is carried out, the radio station is recorded in a visual mode, the radio station number is associated with an actual radio station, and the state of each radio station can be checked through a space monitoring view.

f. The simulator monitoring is carried out, the simulator input is carried out in a visual mode, the association is carried out through the number of the simulator and the actual simulator, and the state of each simulator can be checked through a space monitoring view.

(IV) System service layer

The comprehensive portal mainly provides the functions of subject management, teaching plan management and training resource planning on business.

a. And the subject management provides a subject input function for each professional subject through a tree structure, and can record the teaching subjects of the MNXL center completely.

b. Teaching plan management, through calendar and list view, can enter each stage teaching plan, and teaching plan can be associated with subjects and training resource planning.

c. Training resource planning can be performed through a space view, and places, seats, equipment and software resources used for training are planned.

(V) System Access layer

a. Unified user authentication, which provides unified system login for all systems through a unified authentication platform, and also provides unified user data management functions including user, role, authority and organization management.

b. The desktop application is accessed, and the desktop adopts plug-in and extensible technology, so that the application in the cloud platform can be integrated, and Web applications developed by various manufacturers can be integrated.

2. Application deployment

As shown in fig. 2, the cloud assistant collects system information of the edge PC, and in a training preparation stage, the pilot control software can fetch a software list to deploy and arrange seats to form an application deployment scheme, so as to provide planned deployment and instant deployment.

In order to cooperate to realize the one-key preparation function on the client PC, a cloud assistant tool needs to be developed, and the cloud assistant supports the functions of mirror image downloading, environment data downloading, cloud desktop resource application, mirror image downloading and starting, and provides support for unified application management, and the specific functions are as follows:

(1) Image downloading: and downloading the needed mirror image from the cloud platform according to the service requirement.

(2) And (3) data downloading: according to the service requirement, the required data is downloaded from the cloud platform, the local data and the server data are compared as required, repeated downloading is avoided, and the network pressure is reduced.

(3) Virtual machine image downloading and starting: and downloading the virtual machine image according to the service requirement, and starting the virtual machine image locally.

(4) Cloud desktop starting and connecting: and when the cloud desktop application is entered through the comprehensive portal, the local cloud assistant where the browser is located applies for the corresponding cloud desktop resource, and automatically starts the cloud desktop client to connect with the remote cloud desktop.

(5) The data is returned to the cloud platform: and in the training and withdrawing stage, the data are returned to the cloud platform for storage according to the service requirements.

3. Use flow

As shown in fig. 3, the comprehensive portal aims at solving the unified access problem of different manufacturers and different types of applications, and covers operation and maintenance management, system monitoring, leading application, branching application and environment platform application, including Web application, cloud desktop application, physical machine application and physical virtual desktop application form, and the comprehensive portal provides a unified access entry through the cloud platform in a Web application mode and opens the application in a unified way through the following modes:

(1) The Web application is directly opened through a browser;

Description of the preferred embodiments

Taking the training scene of chemical defense training as an example, the training system and the training system are distributed in a central machine room and more than ten classrooms of the training building, when the training system is organized, the training system and each professional training system are added, and under the condition that the training semi-physical simulator is insufficient, the training system is provided with a desktop simulation training system through a common computer of the desktop training classroom to supplement seats so as to support the full-scale chemical defense simulation training.

As shown in fig. 1, the system has many different space devices to be accessed, including machine room environment monitoring devices, cameras, access control, computers, simulators, simulation radio stations, sound equipment and the like, all devices supporting training are accessed through a device access layer, all devices are recycled, system services are supported in a resource view angle, and an upper service system can further call the training supporting capability of the system according to the secondary development capability of the service system.

As shown in fig. 2, the system provides high-level arrangement capability through a service layer, including arrangement of client computers and client applications, wherein the applications supporting arrangement include cloud desktop applications, physical virtual machine desktop applications, physical machine applications and Web applications, the training environment arrangement can be performed according to the scale of a training person and a training subject in the training preparation process, the application distribution is performed through an application center, and the operation and maintenance management of the client computers is realized through an end management tool.

As shown in fig. 3, from the perspective of system usage, an operation and maintenance user logs into the system through a unified portal to perform one-key preparation of a training environment, wherein: the access to the Web application is the opening of the application by starting a client browser; for cloud desktop client application, the cloud end and a client computer are matched, the cloud end completes preparation of resources, after a cloud desktop virtual machine is created and started, a orchestration system starts the cloud desktop client by receiving a preparation notification, and a user can access the application in the cloud desktop by logging in with a unified account; the physical machine application requires an end management tool to download the virtual machine image from the image center to the local, start the local virtual machine and references in the virtual machine, and provide access for trained personnel; the physical machine application is similar to the physical virtual machine application, and the difference is that the application is lighter and runs on the physical machine, the use experience is better, and the method is suitable for the scene of three-dimensional training software.

Training preparation

The training is a simulation training with maximum standard in annual training, maximum number of participants and higher comprehensive requirement, the preparation of the simulation training is usually carried out 1-2 months in advance, from the steps of designating the planned training subjects, making the training plan, arranging training resources and testing training environment, and finally carrying out test preparation acceptance, the completion of the training preparation is calculated, and the training preparation flow after the application is used is as follows:

1. planning training subjects, wherein a large number of daily training subjects exist in a general subject library, each subject is designed for training points aiming at a training outline, a proper training area (map), training equipment and training tasks are selected, the subjects which are trained daily are required to be optimally combined by the training subjects so as to achieve the aim of training together, the coordination of each trained group is trained mainly, the strain capacity of the trainee under the complex condition (guiding adjustment) is trained, the basic skills of the trainee mastered by theoretical learning are simulated and cooperated in a training scene with high fidelity, and the tactical operation capacity of the trainee is checked and improved.

2. The training program is formulated, the training program is generally formulated in a unified way by a teaching management department, the training program formulated at the educational administration department can be imported into the training operation and maintenance system by the training system through synchronous data, imported data and the like, resource sharing of each training place can be planned more clearly through the training program, unnecessary equipment use conflict is solved, and related matters of training can be informed in time through the reminding function of the system, and implementation of operation and maintenance work such as preparation, implementation and the like of training is promoted.

3. Training resources are arranged, as training program dates gradually approach, actual training environment resources including cloud server computing resources (environment simulation and the like), cloud server data resources (map and model resources), training classrooms, computer resources, guide control rooms, guide control equipment resources and the like are allocated according to training program requirements, system business operation staff track and manage the training process from two dimensions of operation resource bearing and training resource monitoring by understanding a training program tree, combining with an actual training environment, arranging equipment and resources through a model training preparation system in a space view, determining used equipment and resources, arranging client software, carrying out matched monitoring equipment association;

4. after training is carried out through training preparation functions in the training environment, the training environment is complex, and one-key preparation test is needed to be repeated, so that the normal operation of all parts of software and hardware is ensured.

5. After the test acceptance is finished, operation and maintenance personnel repeatedly test, the relevant responsible person is required to accept the test result, and the preparation before training implementation is ensured to meet Shi Xun requirements.

(II) training practice

1. The one-key starting computer starts all the service and client computers on which the system depends according to the training scheme arranged in the training preparation stage, and the technical states of the trained computer and the simulator can be checked through the monitoring center after the system is started;

2. after the hardware equipment is ready to be completed, the training system is started by one key, and the configured client software in the arrangement scheme is started by one key through the monitoring center, so that the starting condition of the training software can be known in time through the starting state returned by the application program and the desktop monitoring picture.

3. Real-time monitoring is carried out, wherein the real-time monitoring comprises machine room environment monitoring, cloud platform resource monitoring, cloud platform service monitoring, trained desktop monitoring and training environment monitoring (cameras);

4. necessary emergency preparation is performed, such as machine room alarm condition processing, software recovery after the client device is matched with replacement, emergency processing is performed by notifying a trained personnel of an abnormal condition through an IP telephone, and the like.

(III) training withdrawal

After the training is finished, a one-key withdrawal function is provided, from the business, a group trainer informs each trained software of the end of training through a guiding control system, each trained software carries out corresponding resource cleaning and data restoring, after the business is cleaned, a cloud platform operation and maintenance system of a model training center can complete software closing, resource returning and computer closing through one key, after a one-key withdrawal instruction is sent, a withdrawal condition can be checked through a monitoring system, and the completion of the withdrawal operation after no abnormality is ensured.

According to the application, a system for preparing and monitoring the modeling and training system by one key is realized based on a cloud platform, hardware resources and applications can be arranged according to requirements through a unified abstract layer according to modeling and training tasks, the use experience of the system is improved, a user only needs to use the system from the view angles of equipment and applications, namely the top, and secondly, the system can effectively schedule the resources through supporting a teaching plan, so that the multi-field modeling and training tasks can reasonably allocate the resources in the prior resources, the resource sharing is supported to the greatest extent, the utilization rate of the existing system resources is improved, the operation and maintenance personnel are released from the low-efficiency and heavy physical work with the support of one-key starting, one-key starting application and the like along with the improvement of the integration level of the system, and more training contents of the modeling and training personnel are reserved, so that the system user can have more energy to engage in training methods and training mode innovation.

The principles and embodiments of the present application have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present application and the core ideas thereof; meanwhile, as those skilled in the art will have variations in the specific embodiments and application scope in accordance with the ideas of the present application, the present description should not be construed as limiting the present application in view of the above.

While the foregoing is directed to embodiments of the present application, other and further details of the application may be had by the present application, it should be understood that the foregoing description is merely illustrative of the present application and that no changes, substitutions, or alterations herein may be made without departing from the spirit and principles of the application.

Claims

1. A method for preparing one-key starting based on a model training of a cloud platform is characterized by comprising the following steps:

(1) The cloud and end resources of the training center are abstracted uniformly, so that the resource arrangement function is provided for training subjects and training plan view angles, and the resource arrangement is carried out or the existing resource arrangement scheme is multiplexed;

(2) At a system access layer, through unified login, the system can access all software of a model training through one account, a computer is started by one key, and the states of a server and a client PC are monitored;

(3) After the hardware equipment is ready to be completed, the training system is started by one key, and the configured client software in the arrangement scheme is started by one key through a monitoring center to monitor the states of the client and the server software;

(5) Training is resumed.

2. The method for preparing one-key start of model training based on a cloud platform according to claim 1, wherein the step (1) specifically comprises:

3. The method for preparing one-key start-up based on model training of cloud platform as set forth in claim 1, wherein,

4. The cloud platform-based model training preparation one-key starting method according to claim 3, wherein the cloud assistant has the following specific functions:

s1, mirror image downloading: downloading a required mirror image from the cloud platform according to service requirements;

s2, data downloading: according to the service requirement, downloading the required data from the cloud platform, comparing the local data with the server data according to the requirement, avoiding repeated downloading, and reducing the network pressure;

s3, downloading and starting the virtual machine image: downloading the virtual machine image according to the service requirement, and starting the virtual machine image locally;

s4, cloud desktop starting and connection: when the cloud desktop application is entered through the comprehensive portal, a local cloud assistant where the browser is located applies for corresponding cloud desktop resources, and automatically starts a cloud desktop client to connect with a remote cloud desktop;

s5, data are returned to the cloud platform: and in the training and withdrawing stage, the data are returned to the cloud platform for storage according to the service requirements.

5. The method for preparing one-key start of model training based on a cloud platform according to claim 1, wherein the step (1) specifically comprises:

6. The cloud platform-based training preparation one-key starting method is characterized in that training resource planning can be performed through space view, and resources such as sites, seats, equipment and software for training are planned.

7. The method for preparing one-key start-up based on model training of cloud platform as set forth in claim 1, wherein,

8. The cloud platform based module training preparation one-key starting method according to claim 1, wherein application opening is uniformly performed by:

1) The Web application is directly opened through a browser;

2) The cloud desktop application clicks on the browser, simultaneously triggers a background to prepare cloud desktop resources, simultaneously informs a user desktop of the current cloud desktop preparation state through a cloud assistant, and automatically logs in the prepared cloud desktop through carrying user login information by the cloud assistant after the cloud desktop preparation is completed;

3) The physical machine application clicks on the browser and triggers a user desktop cloud assistant at the same time, if the local software does not exist, the software pulling is carried out, the software pulling state is displayed, if the local software exists, a local cache is used, and software starting is carried out after the software preparation is completed;

4) And the physical virtual desktop application clicks on the browser and triggers a user desktop cloud assistant, if the local virtual machine mirror image does not exist, mirror image pulling is performed, the mirror image pulling state is displayed, if the local virtual machine mirror image exists, a local cache is used, and after the virtual machine mirror image preparation is completed, the local KVM is used for starting the mirror image.

9. The method for preparing one-key start-up based on model training of cloud platform as set forth in claim 1, wherein,