Cloud-edge Collaboration Based Methods in Improving the Safety Inspection Efficiency on Electric Operation Scenes

The safety in electric operation scenes directly influences social economic development and stability. Safety supervision and inspection is one of the most important links in the electricity generation process. The previous mode of executing supervision and inspection in advance and analyzing and reporting afterwards cannot satisfy the requirements of the modern management system, which requests beforehand prevention, in-process control, post review, and analysis. The supervision and inspection efficiency is low. Therefore, this paper proposed three innovative methods to increase the supervision and inspection efficiency of the electric operation. First, we proposed an operation demonstration method on electric equipment that integrates with Augmented Reality (AR). It can greatly increase the operating accuracy and decrease the probability of accidents. Then, a smart, mobile, and cloud-edge collaborated method is proposed, aiming at increasing the availability and effectiveness of safety supervision and inspection. Last, we adopted technical means of integrative recognition to increase the efficiency of obtaining evidence on inspection.


Introduction
National grid company takes safety as the most fundamental and most important cause.In the process of power generation, safety supervision and inspection are one of the most crucial duties.Prior intervention, in-process leak filling, and afterward summary and evaluation of illegal and violation operations contribute to the safe production capacity.
In recent years, power demand has been increasing, along with a higher ratio of renewable energy and more frequent natural disaster risks.The stability and safety requirement of power grids is getting higher and higher.The operation mode of power grids also changes along.Therefore, the previous supervision and inspection mode on working sites cannot effectively support the modern safety management system.Moreover, the existing automation system cannot cater to power grid scale growth either.
On the one hand, the current safety supervision and inspection on power construction sites mainly rely on manual work.In this process, it costs much time and effort on dealing with the data acquired.Data and reports cannot be submitted on time.As the construction and operation sites continue to grow and become more complex, especially for those small and dispersive sites, the task of supervision and inspection has become more arduous.
On the other hand, the current equipment used for supervision and inspection is relatively backward.The interaction between supervision equipment and the cloud platform is weak.The platform cannot grasp the real-time inspection content and key points on the scenes, which indirectly lowers the supervision and inspection efficiency [1] .

Current pain spots and necessity
In this section, we will first talk about the pain spots of supervision and inspection on electric operation scenes.There are three major points.Firstly, the supervision and inspection work is highly experience dependent.Secondly, the inspection work is time and energy-consuming.A variety of illegal operations lead to difficulty in forensics.The last is that site data cannot interflow with the cloud platform on time.Various operation types and complicated scenes also raise the bar for supervision and inspection.The application of intelligent identification tools is limited.
The three methods proposed in this paper can address the above problems.Considering the high experience-dependent characteristics of the work, process digitization and making auxiliary equipment more portable and intelligent is crucial for increasing supervision and inspection efficiency.And unmanned identification applications relying on machine vision cannot be realized in a short time.
In addition, it is also necessary to develop a set of smart, digital, and mobile carriers for supervision and inspection.Such carriers should integrate intelligent management terminals.Taking advantage of 5G, BDS and AI technologies, mobile and noninductive inspection and intelligent identification of illegal operations can be realized [6] .In such a way, the dependency on human resources and experience can be decreased, and efficiency, coverage, width, and depth of supervision and inspection can be enhanced.In the next section, we'll give a detailed description and demonstration of the three methods mentioned above.

Smart digital mobile supervision and inspection method based on cloud-edge collaboration
In terms of low-time effectiveness, flexibility, and efficiency in current supervision and inspection work, this chapter proposed a Smart digital mobile supervision and inspection method based on cloud-edge collaboration.The guideline for this method is shown below.
First, standard procedures of supervision and inspection in various operation scenes are summarized.The inspection key points for scenarios are sorted out, such as power transition, transformation, distribution, capital construction, and marketing.Second, considering the big bandwidth and low time delay advantages of 5G technology, the application of 5G UAV (Unmanned Aerial Vehicle), and 5G action recognition on grid powerline inspection and maintenance are sorted out.Finally, a multidimension framework for mobile safety supervision and inspection mode is constructed, including a physical layer, perception layer, network layer, platform layer, and application layer.
This chapter gives the architecture of the mobile safety supervision and inspection system, as demonstrated in Figure 1.In the physical layer, a unified communication protocol is adopted, which collects UAVs, smart helmets, body-worn cameras, and other safety management terminals.In such a way, efficient access to perception data of inspection can be achieved.
Data-collecting terminal equipment is mounted on the user-side carrier, i.e., the inspection vehicle, to collect multiple types of data and information.In the network layer, also known as the "Tube" in the "Cloud-Tube-Edge-End" infrastructure, a combination of 4G, 5G, and BDS communication technologies is used.
The hardware scheme of the mobile supervision and inspection system is shown in Figure 2. Vehiclemounted edge calculation devices should be first developed.Then an integrated safety inspection system using vehicles as carriers is designed, including various clients such as 5G UAVs, intelligent helmets, and 5G body-worn cameras.Finally, inspection applications, such as illegal operation identification, action recognition, and comprehensive analysis should also be developed.

Smart digital mobile supervision and inspection method based on cloud-edge collaboration
Electric equipment operation faces the following typical problems: environment complexity, low efficiency, the faultiness of operation data, no online closed loop for operation data, and so on.To address the above problems, this chapter brings up an operation demonstration method for power equipment that applies AR technology.This method adopts AR technology and 5G communication technology, turning the content in digital work order tickets and operation tickets into AR objects that seamlessly unite the physical operation equipment.It can help operators quickly locate the specified equipment, and comply with the instructions of AR objects to orderly, correctly, and efficiently finish the operation.
After the operation, the handheld demonstration system will automatically upload the operation content and process, thus closing the online loop for operation data.The workflow of the handheld operation demonstration system for electric equipment based on AR and the use process are shown below in Figure 3.To begin with, the operator holds the handheld device towards the specified electric equipment.The handheld device confirms the information of the equipment by image identification.Afterwards, the handheld device acquires AR dynamic objects of operation procedure generated by the AR demonstration system and shows them on the screen in real-time [7] .That is, the operator can get a clear visual view of the object, and how to handle and operate the equipment as well.The operator then successively finishes the operation according to the AR dynamic objects.
When the operation is finished, the handheld device will automatically upload process data of the operation via dedicated 5G wireless access to the cloud platform and close the data loop.The operation data includes time, location data, operation images, operation objects and processes, warning messages, and so on.Meantime, the operation results will be sent to the work order signer to close the business loop.
Compared with current methods, highlight advantages of the handheld power equipment operation demonstration system based on AR include 3 aspects: visualizing the operation process by AR demonstration system, increasing operation efficiency, and decreasing operation error probability.

Inspection forensic system for electric operation scenes
Forensic is one of the important steps in the safety inspection of electric operation scenes.Traditionally, inspectors use their phones to collect evidence of operation faults, including management faults, action faults, equipment faults, and so on.In such a process, it takes quite a long time to collect evidence, label fault details, collate data and write reports.
Forensic is so time-consuming that up to 80% of the time is consumed in this process, not to mention complicated scenes and tedious inspection processes.It is the most influential factor that the efficiency of inspection work cannot be raised.Therefore, this chapter brings up a novel forensic method and device for electric operation scenes.The device is used for assisting inspectors to identify operation faults, verifying working stuff and equipment, recording unmatched information, and collecting image evidence.
The hardware of the forensic client consists of a data storage unit, an image processing unit, a BDS, a front-facing camera, and hardware including a power source, an 5G wireless communication module, and a charging module.
The software of the forensic client includes a fault library, which contains content operation faults, equipment faults, and management faults.The automatic forensic system consists of an image identification module, action recognition module, face recognition module, character identification module, and scanning module.Other software includes an automatic voice recognition module, an inspection report generation application, and a wireless 5G communication module.The process of using this forensic device to conduct an inspection is shown in Figure 4.  Inspectors take the handheld device to the operation site.Then the forensic client will automatically obtain comprehensive information about the operation site online.The automatic forensic application of the handheld device can recognize operation faults and illegal actions via image identification functionality.Afterwards, the client will add detailed violation information to the images acquired through the front camera, including time, location, operation content, and violation terms.As for fault operations that cannot be identified by the device, inspectors can manually take pictures and add violation information.
Personnel management is crucial for operation scenes, because unqualified crews may be untrained staff.So, to verify the identities of the operators at the site, inspectors can use the front camera to do face recognition and compare it with the information on the cloud in real-time.If the information does not match, it means the staff is not in the operation scheme, which violates the management regulations.The client will also automatically record the violation information.To verify the information on equipment on site, the method requires that all the equipment should put up a QR code or bar code that contains the necessary information.Inspectors scan the QR code or bar code to obtain the information and compare it with log-on details in the background.When the forensic process is done, the client will automatically generate a forensic report and upload it to the cloud platform through 5G wireless access.
By using the proposed inspection forensic system and client, the forensic efficiency can be greatly improved and the forensic time will be shortened by 50% at least.

Conclusion
This paper proposed three innovative and feasible methods that can improve the efficiency of supervision and inspection on electric operation sites.The first method merges edge computation and cloud platform, and forms a smart, digital and mobile safety supervision system based on cloud-edge collaboration, which can greatly improve the flexibility and time efficiency of the work.Then we put forward a system that combines AR technology with a portable device to demonstrate the detailed operation process of electric equipment, which can decrease the probability of operation faults.Last, to improve the forensic efficiency at the operation site when inspecting, a fusion recognition method that helps recognize faults and collect evidence is proposed.It is supposed to decrease the forensic duration by 50%.The utilization of the above three methods can help the supervision and inspection team improve efficiency, and further promote safer power production.

Figure 1 .
Figure 1.Framework of the mobile supervision and inspection system based on cloud-edge collaboration

Figure 2 .
Figure 2. Hardware scheme for mobile safety inspection system using smart vehicles as carriers

Figure 3 .
Figure 3. Workflow of handheld power equipment operation demonstration system based on AR

Figure 4 .
Figure 4. Forensic process using device based on fusion recognition