Design and Implementation of Adaptive Protection Strategy for 0.4kV Safety Anti Electric Shock Maintenance Power Box in Smart Grid
Online veröffentlicht: 17. März 2025
Eingereicht: 17. Okt. 2024
Akzeptiert: 06. Feb. 2025
DOI: https://doi.org/10.2478/amns-2025-0316
Schlüsselwörter
© 2025 Yong Wang et al., published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.
In today’s society, the rapid development of market economy, electricity is one of the most widely used energy sources, involving all areas of social life, whether it is transportation, production sector, household appliances, or scientific and technological aspects, can not be separated from the supporting role of electric energy [1-2]. Of course, in every corner of social development, the shadow of electricity is ubiquitous and ubiquitous, but in the provision of electric energy at the same time, the safety issue must not be ignored, a little inattention, there will be electrocution accidents [3-4]. As we all know, the occurrence of power accidents, only a few belong to the unforeseen and irresistible, a large number of power accidents can be completely prevented, and all this needs to be done from the power maintenance work. In view of this, in the electric power maintenance work, we must put the prevention of electrocution accidents in the first place, set up the “safety first”, “prevention-oriented” concept, and actively take effective preventive measures [5-6].
In the current electric power maintenance work, the design of most electric power equipment is not reasonable enough. For example, for the substation equipment, in terms of design, manufacturing and installation, due to the unscientific design program of individual substation equipment, it directly leads to the installation of electric power equipment with large safety hazards [7-9]. Of course, in this case, when the electric power maintenance personnel carry out the operation, it is very easy to cause the danger of personal electrocution, and the chances of such power accidents are very high. In addition in the vicinity of energized equipment construction or there is a power outage, but many of the energized equipment with insufficient warning, increasing the risk of personal electrocution. For example, for some of the substation’s energized equipment, once the warning is not in place, it is easy to make the power maintenance personnel mistakenly man the energized intervals, or worse, will directly walk on the energized equipment [10-12]. There is no doubt that the consequences of mistakenly on the energized equipment will be unimaginable, basically will result in the tragedy of personnel electrocution death. And the degree of automation of power equipment is not high, power maintenance work to meet the basic needs of modern power development, it is difficult to achieve the goal of complete unmanned so that there are often communication interruptions, monitors can not remote control of on-site equipment, maintenance personnel tired of running for their lives and other undesirable phenomena coupled with the monitoring equipment mistakenly sent interference signals are also interfering with the judgment of the monitor to a certain extent, and if these signals are gradually evolved into a major accident information, then it will be the most important thing. If these signals gradually evolve into major accident information, it will exacerbate the expansion of power accidents [13-14]. In addition, in the operation process of electrical maintenance, many maintenance personnel are not clear about the structure and basic principles of electric equipment, not capable of coping with complex electrical operations, it is inevitable to carry out the wrong operation, some maintenance personnel are greedy for saving time, lack of responsibility and safety awareness, even if they found the equipment defects but indifferent, do not report the refurbishment in a timely manner, so as to omit the item skipping, which is prone to human-caused power accidents [15-16].
On the other hand, standardizing safe work behavior and developing good work habits are the personnel quality and work requirements that must be possessed to prevent personal electric shock accidents. From the personal electric shock accidents that have occurred and a number of personal electric shock accidents that have occurred in the system, it can be seen that the high incidence of personal electric shock accidents is in the 35kV and the following indoor power distribution unit overhaul work. In addition to taking some targeted measures from the safety technology means and safety management, the most important thing is to develop personal good working habits from ensuring their own safety [17-18].
In the electrocution accident of the service power box, it is one of the important reasons that the personnel who are not qualified for authorization open the service power box, thus leading to electrocution. Combined with the application case of blockchain technology in smart grid, this paper adopts the strategy of establishing the security identification model of the service power box to realize the adaptive protection of the service power box. By configuring the range of electric power maintenance personnel allowed to open each type of maintenance power box, the opening authority of the maintenance power box, the authority of the work ticket and the operation ticket, a unified authorization system for maintenance power box safety is constructed. Install cameras and extract the facial information of on-site personnel. Remotely obtain the work ticket information for the maintenance power box through an encrypted mobile network. Use an encrypted mobile network to remotely obtain the operation ticket information for the maintenance power box to verify the personnel responsible for maintaining it and the permitted time range for opening it. Authenticate the authority of the on-site personnel through blockchain technology to realize mutual trust and authentication between the personnel and the work ticket, the operation ticket and the safety lock of the power box, so as to reduce the occurrence of electrocution incidents.
Electrical maintenance of substation equipment in the smart grid is an important measure and method to test the technical indicators of substation equipment, due to its technical and professional nature. The high overhaul voltage characteristic of the overhaul process leads to a significant increase in the danger and destructiveness of safety accidents. In the substation for routine maintenance and test work, often use the maintenance power box as an external power supply point. Summarize the various types of electrical maintenance accidents found that more than 60% of the accidents, is due to maintenance personnel in the case of not having the authorization to open the maintenance power box, maintenance personnel forget to turn off the power supply and other factors, so that the maintenance equipment or the maintenance equipment charged, so that the maintenance personnel electrocution, resulting in casualties. “Zero accident safety production” is the eternal pursuit of electric power enterprises, in the electrical maintenance work, improve the safety factor of electrical maintenance power supply, and effectively protect the personal safety of maintenance personnel is very significant. Therefore, in order to ensure the personal safety of maintenance personnel in a wide range of electrical equipment maintenance work, there is a need for a safe anti-electrocution maintenance power box protection strategy.
Blockchain is a distributed database that forms a data structure called a block by arranging transaction records in chronological order. Each block contains the hash value of the previous block, thus forming a tamper-proof chain structure, which makes blockchain technology highly secure and reliable. Blockchain technology is characterized by decentralization, tamperability, and anonymity, and is widely used in smart grids.
Blockchain-based power transaction system. With the development of the power industry, power transactions become more and more complex. The traditional power transaction system has some problems, such as insufficient trust mechanisms, lack of transparency, and security issues. Blockchain-based power transaction system is a new type of power transaction model, which utilizes the decentralization, inerrancy, anonymity and other characteristics of blockchain technology to record all aspects of power transactions on the blockchain, which achieves a more secure, efficient, transparent and fair power transaction [19].
Blockchain-based energy management system. Blockchain-based energy management system can record the information of energy production, consumption and transaction on the blockchain, and utilize the characteristics of blockchain technology, such as decentralization, tamperability and smart contract, to interact energy suppliers, system administrators, large-scale energy-purchasing users and load aggregators on an equal footing as different counterparts in the blockchain network, and to realize the distributed management, transaction and sharing of energy, so as to realize the whole life cycle management of energy [20].
The framework of the maintenance power box security identification method based on blockchain technology described in this paper mainly includes six parts, including the maintenance power box security unified authorization system, on-site personnel face information extraction, work ticket information extraction, operation ticket information extraction, maintenance power box authority authentication, and mutual trust authentication of maintenance power box. The maintenance power box security identification framework can be seen in Figure 1.
Repair power box security identification framework diagram
Inspection power box security unified authorization system is the entire inspection power box security identification model account and power maintenance user authority control basis. Power supply security unified authorization system includes power maintenance personnel account management, maintenance power box identification identity authentication, power maintenance personnel authorization and authority control. The maintenance power box security authorization system is divided into maintenance power box authority, operator authority, work ticket authority, and operation ticket authority.
Maintenance power box authority
Power maintenance site maintenance power box according to the location of the distribution line can be divided into a maintenance power box, two maintenance power box and three maintenance power box. Among them: the first level of maintenance power box installed in the power transformer on the side of the low-voltage outlets, the second level of maintenance power box installed in the low-voltage branch line side, the third level of maintenance power box installed in the maintenance of the end of the side. The maintenance power box can be divided into 0.4kV and 0.22kV maintenance power boxes depending on the voltage level. Among them, the first and second level maintenance power boxes are 0.4kV, and the third level maintenance power box is 0.22kV.
In the maintenance power box authority settings, the main configuration allows each type of maintenance power box to open the range of power maintenance personnel. The range of maintenance power box authority is as follows:
Where:
Operator rights
Inspection power box operator rights are divided into three categories: electric power maintenance site management personnel, electric power maintenance site operators, and electric power maintenance site quality supervision personnel. Among them, the power maintenance site managers are responsible for the maintenance site and have the authority to audit the operation ticket and work ticket, and the power maintenance site operators and power maintenance site quality supervisors have the authority to open the maintenance power box. Personnel authority collection
Where:
work ticket, operation ticket authority
The work ticket and operation ticket authority mainly controls the power maintenance personnel to work on the maintenance power box opening range and allow the start and end time of the opening box, so as to avoid the power maintenance personnel to open the maintenance power box at will.
When a camera is installed at the locking device of the overhaul power box to extract the face information of the field personnel, the face photo of the power overhaul personnel will be acquired and the face features of the field personnel will be extracted by the local feature region algorithm so as to reduce the amount of data transmission.
The local feature region algorithm is an important process for extracting and converting the security features of the face information of the overhaul power box. It converts face information into feature data vectors and has been widely used in face information extraction [21]. The extracted face feature information is:
Where:
The operator of the power maintenance site fills in the work ticket and operation ticket. The power maintenance site manager with administrative authority approves the work ticket and operation ticket. After approval, a process order containing information on the opening of the power box is created. The maintenance power box security authentication module remotely obtains the work ticket and operation ticket information of the maintenance power box through the encrypted mobile network, and determines the permissible opening time range of the safety box and the permissible information of the power maintenance personnel.
In the maintenance power box authority authentication: firstly, the maintenance power box collects the face feature information of the on-site maintenance personnel, secondly, the maintenance power box will initiate the authentication request to the security boundary, and finally, the maintenance power box adopts the blockchain technology to realize the authority authentication of the on-site maintenance personnel. The authentication data
Where:
In the access power box mutual trust authentication session, the blockchain receives information about the staff member’s authority and authenticates this information with the work ticket, the operation ticket and the access power box safety lock. Mutual trust authentication information
Where:
It mainly includes the power supply input part, the output meter connection part, the control processing part and the wiring board part. The working principle diagram is shown in Figure 2. Achievable function: Even if the terminal strip is immersed in water, the leakage current does not flow to the outside, but only flows between the connecting end of the terminal strip and the leakage-proof conductor, which can effectively prevent leakage and prevent electric shock when the human body is exposed to water.
The working principle diagram
Take the 0.4kV service power box as an example to analyze the recognition effect of the safety recognition model constructed in this paper. The face tilt angle experiment uses the CAS-PEAL multi-pose face library to extract 120 people with 18 images each for training. In order to further illustrate the performance of the system, the face recognition rate of this model was counted, and two other multi-pose face recognition algorithms were compared, respectively LLR algorithm and Gaussian regression algorithm, and the face recognition results are shown in Table 1.
Face recognition results
| Face rotation Angle | Our model(%) | LLR(%) | Gaussian regression(%) |
|---|---|---|---|
| 25° | 97.36 | 98.73 | |
| 40° | 89.83 | 90.27 | |
| 55° | 85.25 | 88.60 | |
| 70° | 82.53 | 87.37 |
From the results, it can be seen that the recognition rate of this paper’s algorithm between the rotation angles [25°, 70°] is overall better than similar recognition methods, especially in the case of rotation angles greater than 40°. As the rotation angle increases, the recognition rates of all three methods gradually decrease, mainly due to the classification difficulties caused by the gradual decrease of the face features contained in the side face. For faces with a rotation angle greater than 70°, it is difficult to localize the feature points in this model, and pose estimation is not possible, and the recognition rate decreases dramatically, and future work will be the research of recognition methods for faces with large rotational angles.
Access Power Box Security Recognition Accuracy is the accuracy of the access power box security lock in recognizing users. This index is calculated as the ratio of the number of successful lock openings to the total number of recognitions. This experiment utilizes the ORL dataset as the experimental dataset and selects various classical algorithms and the overhaul power box security recognition model for comparison experiments. The data for the safety recognition experiment results are shown in Table 2.
Safety identification results
| Sample size/group | Accuracy(%) | |||||
|---|---|---|---|---|---|---|
| Our method | PCA-SVM | LBP | PCA-LDA-SVM | LBP+MB-LBP | LBP+LDRC-Fisher | |
| 500 | 99.23 | 97.37 | 96.15 | 92.53 | 90.62 | 88.72 |
| 1000 | 99.03 | 97.56 | 97.51 | 92.38 | 90.38 | 88.72 |
| 2000 | 99.03 | 97.41 | 96.93 | 92.87 | 89.50 | 89.98 |
| 3000 | 98.88 | 97.22 | 96.54 | 91.79 | 88.52 | 88.03 |
| 4000 | 98.73 | 97.61 | 97.47 | 92.09 | 89.01 | 89.40 |
| 5000 | 98.93 | 96.83 | 96.34 | 91.79 | 89.30 | 87.74 |
From the experimental data results comparison analysis of this paper’s method and other types of algorithms, through the experimental test on the ORL face dataset, this paper’s model in the case of different number of samples, the achieved security recognition accuracy are more than several other algorithms, its average recognition rate is 98.97%. While the average security recognition rates of PCA-SVM, LBP, PCA-LDA-SVM, LBP+MB-LBP and LBP+LDRC-Fisher for servicing power boxes are 97.33%, 96.82%, 92.24%, 89.56% and 88.77%, respectively. In addition, the safety recognition rate of this paper’s model performs more stably under different sample sizes, and the standard deviation of the computed samples is 0.15, which is 40.5% less than that of the PCA-SVM algorithm.
In order to more intuitively compare the recognition accuracy achieved by various types of algorithms under different sample sizes, the accuracy of various types of algorithms is presented in a line graph with a more obvious comparison effect, and the comparison of the recognition accuracy of various types of safety recognition methods is shown in Figure 3. It can be seen that the recognition accuracy of the safety recognition model of the service power box proposed in this paper is consistently and stably higher than that of the comparison methods under different sample sizes. It can be seen that this paper’s maintenance power box safety recognition model to ensure a higher recognition rate at the same time, recognition effect fluctuation is smaller, recognition performance is more stable, to help prevent unrelated personnel to open the maintenance power box, reduce the occurrence of electric shock.
Safety recognition accuracy of different sample sizes
Traditional face recognition algorithms need to be effectively trained on a sufficient number of face images for a given identity before they can correctly recognize that identity. When the system is faced with a stranger who has never appeared in the sample library, it often fails to recognize him or her effectively, resulting in a decrease in the security and reliability of the system. The number of samples of stranger faces is theoretically infinite, so it is impossible to provide effective samples of stranger faces. When the field personnel face information aliasing method is promoted in the practical application of the adaptive protection of the maintenance power box, it is necessary to test the model’s recognition accuracy of the stranger’s face.
This experiment uses the correct rate, omission rate, and detection rate as the testing indexes, where: correct rate = number of strangers correctly identified ÷ total number of stranger samples, omission rate = number of strangers identified as authorized personnel ÷ total number of stranger samples, and detection rate = number of strangers correctly identified ÷ total number of samples identified as strangers. This experiment uses the ORL dataset as the experimental dataset to simulate the stranger recognition scenario, and the experimental results are shown in Table 3.
Recognition results of inexperienced faces
| Methods | Accuracy | Miss rate | Detection rate |
|---|---|---|---|
| Eigenfaces | 81.2% | 18.8% | 95.7% |
| Fisherfaces | 75.8% | 24.2% | 96.2% |
| LBPH | 82.7% | 17.3% | 96.5% |
| Our method | 95.3% | 4.7% | 98.9% |
Three types of face recognition algorithms provided in the OpenCV image processing library: Eigenfaces, Fisherfaces, and LBPH are chosen as the comparison algorithms for this experiment. From the experimental results, it can be seen that compared to the three types of face recognition algorithms in OpenCV, the model designed in this paper is better in all indicators. The correct recognition rate of the stranger’s face reaches 95.3%, the leakage rate is controlled at 4.7%, and the correct detection rate is 98.9%. From the 98.9% correct detection rate indicates that when the system recognizes a stranger, its judgment result is very likely to be correct, further illustrating the effectiveness of the model in this paper in the safety recognition of the maintenance power box.
To evaluate the effectiveness and time-consumption of this paper’s method and the traditional scanning technique for secure recognition, this paper chose to conduct a comparison experiment on the ORL face database. Table 4 shows the experimental results of the two secure recognition methods, the horizontal rows of the table indicate the number of recognition samples are 1000, 2000, 3000, 4000 and 5000 images respectively. From Table 4, it can be seen that: when the number of security identification samples is the same, the recognition rate of this paper’s method is better than the traditional code-sweeping technology, and in the recognition time of this paper’s method takes significantly less time than the traditional code-sweeping technology, and with the increase of the number of samples, the time used by the traditional method also increases substantially, while the time required by this paper’s method is relatively stable, which also argues for the superiority of blockchain technology in the realization of the overhauling of the power supply box nature of the self-adaptive protection. Adaptive Protection.
The comparison of sample recognition accuracy and time consuming
| Subject | Method/sample size | 1000 | 2000 | 3000 | 4000 | 5000 |
|---|---|---|---|---|---|---|
| Accuracy (%) | Code scanning | 97.82 | 98.21 | 98.11 | 97.69 | 97.62 |
| Ours | 99.03 | 99.03 | 98.88 | 98.73 | 98.93 | |
| Time (ms) | Code scanning | 142 | 655 | 1132 | 2077 | 3352 |
| Ours | 38 | 55 | 72 | 80 | 87 |
The raw experimental data is further processed and analyzed, and Figure 4 shows a graphical comparison of the two methods of safety recognition of the access power box. As can be seen from the figure: with the increase in the number of samples, the recognition rate of both the sweep code recognition and face recognition rate remains relatively stable, while at the overall level, compared with the sweep code method, this paper’s method has a higher recognition rate. The result of the change of recognition time of the two methods with the increase of samples can be seen intuitively: sweep code security recognition in the recognition of the number of samples is 5000, the recognition time has reached 3352ms, while the method in this paper only used 87ms, indicating that the performance of the security recognition method proposed in this paper is more stable and more efficient than the traditional security recognition method.
Comparison of safety identification methods
For identity security recognition, the more common types of attacks include brute force attack, man-in-the-middle attack, replay attack, and denial-of-service attack, and the following will introduce the security of the security recognition model proposed in this paper, which is a fusion of face recognition and blockchain technology, in the face of the above four kinds of risk problems.
In this paper, it is more difficult to crack through the comparison of two-fold authentication factors of user face information and and user identity information, compared with the single-factor authentication method. Even if the user has stolen the face image for malicious login, it is still necessary to verify the identity signature comparison of the hash value of the user information, if you want to violently decrypt the user identity information, you need to traverse (assuming a total of m public keys) to find the corresponding user’s public key to decrypt the signature, due to the hash value can not be cracked in the reverse direction, so only through the Merkle algorithm calculates an equal signature to confirm that the two hash value of the user’s information. In the calculation process, if each message has n possibilities, the difficulty of violently breaking the signature information is m*n^2. And in the case of not obtaining the private key, it is extremely difficult to forge the signature, even if the digital signature is forged, the hash value obtained after decrypting the hash value through the public key does not match the hash value of the original data, and it is still impossible to pass the verification.
Man-in-the-middle attacks are typically caused by DNS spoofing and other methods of interfering between the two sides of the communication, and attacks occur by stealing, forging, and spreading false data. The decentralized identity security recognition mechanism proposed in this paper can effectively avoid this risk for the following reasons: first, the decentralized authentication network relies on each consensus node for information transmission, and in the process of information broadcasting, the man-in-the-middle can not know the exact source of the information, and can not obtain the address of the destination host. Even if the intermediary obtains the information, since the information is transmitted through asymmetric encryption, it is impossible to decrypt the information without access to the private key of the recipient.
Replay attack mainly obtains information such as authentication data that has been sent through listening or other methods, and sends it to the authentication server again. The security identification mechanism proposed in this paper can effectively resist replay attacks for the following reasons: the nodes in the blockchain network designed in this paper will hash the information in the block when updating the block, which will include the calculation of timestamps, due to the timeliness of the timestamps, the nodes in the blockchain network can accurately determine the validity of the message by judging whether the timestamps are valid or not, so as to effectively resist replay attacks.
Denial-of-service attacks are mainly used to prevent the target machine from providing normal services by occupying network bandwidth and making illegal connections. The security identification scheme designed in this paper can effectively resist denial-of-service attacks for the following reasons: the network nodes in this scheme are functionally equivalent and are able to provide complete services, and an attacker needs to attack more than 50% of the nodes to have a fundamental impact on the authentication results. It would also be extremely costly to carry out this attack operation.
This paper combines face recognition and blockchain technology to construct a safety recognition model for the maintenance power box, which aims to prevent unrelated personnel from opening the maintenance power box by mistake, which will reduce electrocution accidents. The face recognition module of the model maintains a recognition rate of more than 90% at a rotation angle between 25° and 70°. The average recognition rate of the model in this paper is 98.97%, with stable recognition performance under different recognition sample sizes. The correct recognition rate for stranger faces in the security recognition model is 95.3%, the missed detection rate is 4.7%, and the correct detection rate is 98.9%. Compared with the traditional code-sweeping identification method, the safety identification technology of the overhaul power box based on blockchain technology not only has better identification accuracy, but also reduces the identification time significantly. The Sweep code security identification for the number of samples is 5000. The recognition time has reached 3352ms, while this paper’s method takes only 87ms. In addition, blockchain technology also has advantages in anti-violent cracking, anti-middleman attacks, and anti-replay attacks, and achieves better security.
Northeast Branch of State Grid Corporation of China Science and Technology Project Funding: Research and application of .4kV safety anti-electric shock overhaul power box (Project Number: 522339240004).