Transmission line detection solution based on the Internet of Things

1 Introduction

On-site environmental monitoring of transmission lines relies on sensors directly installed on the transmission lines that can record and characterize the operating status of equipment in real time to achieve online measurement, diagnosis and maintenance of transmission lines. It is very important for the safe operation of high-voltage and ultra-high-voltage power grids. With the rapid development of micro-electromechanical systems, system-on-chips, wireless communications and low-power embedded technologies, the IoT has provided an all-weather, low-cost, high-reliability and high-redundancy solution for on-site environmental monitoring of transmission lines. Starting from the existing transmission line monitoring system, this paper designs an on-site environmental monitoring solution based on IoT by analyzing the key technologies of IoT.

2 Problems with the existing monitoring and early warning system

Electric power workers at home and abroad have conducted a lot of research on environmental monitoring of transmission lines. In the early days, manual inspection was mainly used to monitor the icing of power facilities. With the development of computer networks and communication technology, the literature developed a computer monitoring system for power facilities using the power communication network; a company introduced GPRS (GSM/CDMA) technology and video technology into the monitoring of transmission facilities and developed a real-time monitoring system for icing of overhead transmission lines. The literature introduced the disaster monitoring system for transmission lines and the online monitoring system for icing. These devices have achieved certain results in the actual field, but there are still the following problems: ① Manual monitoring methods require a lot of manpower and material resources, and cannot achieve 24-hour real-time observation. At the same time, due to the wide wiring range and the harsh geographical environment of some wiring areas, it is impossible to achieve full-range monitoring; ② Ice and snow disasters cause power lines to collapse and break, and at the same time, a large number of communication optical cables are broken. The public communication network and the power communication network have been interrupted to varying degrees, and the monitoring data cannot be reliably sent to the monitoring center; ③ Affected by factors such as region, climate, and terrain, specific monitoring areas require specific alarm strategies, and targeted monitoring data accumulation and strategy improvement are required.

3 Overview of the Internet of Things

Figure 1 shows the structure of the IOT system. The sensor nodes have the functions of perception, calculation and communication. Each node can collect environmental data (such as temperature, humidity, wind speed, vibration frequency and amplitude, etc.), communicate with each other using wireless multi-hop mode, and process the collected data in the network according to the application and system requirements. The aggregation node collects the information collected and processed by the sensor network and delivers it to the user through the Internet or satellite. The user is the receiver and user of the perceived information, which can be a person, a computer or other device.

As a component element of IOT, sensor nodes are generally composed of four basic components, as shown in Figure 2.

The sensing unit senses the environment and generates sensing data, usually composed of a group of miniaturized sensor devices. The processing unit (usually with built-in memory) processes the sensor data and controls the node, so that it can cooperate with other nodes to complete the assigned sensing tasks. Generally, a low-power microprocessor is used, such as the MICA2 Mote system, which uses a 7.37 MHz 8-bit ATMega12 8L microprocessor, has 128 kB program flash memory, 4 kB SRAM, and consumes 16.5 mW. It usually runs on TinyOS, MANTIS and other miniaturized operating systems specially customized for IOT. The transceiver unit ensures that the nodes communicate with each other. IOT generally believes that short-range wireless low-power communication technology is more suitable. At present, with the popularization of ZigBee (IEEE802.15.4) technology, IOT has widely adopted ZigBee devices. The energy unit provides the energy required for the normal operation of the node. Since IOT usually works in an unattended state, the network life depends on the amount of node energy, so energy saving is an important factor in IOT design.

[page]4 Hardware selection for monitoring system

At present, there are many hardware platforms for IOT nodes at home and abroad. Typical nodes include Mica series, Sensoria WINS, Toles, μAMPS series, XYZnode, Zabranet, etc. In fact, the main difference between each platform is the use of different processors, wireless communication protocols and different application-related sensors. Among them, the Mica series nodes are more mature and widely used.

The microprocessor chip of the Micaz node uses Atmega128. The Micaz 51-pin expansion interface can connect analog input, digital I/O, I2C, SPI interface and UART interface. The communication module uses the CC2420 chip. This chip is the earliest communication chip that supports Zigbee communication technology. The carrier frequency is 2.4 GHz, the data transmission rate is up to 250 kbps, and the communication distance is 60 to 150 m, which is more suitable for indoor applications. The data acquisition module uses the ADXL202JE accelerometer, which can simultaneously collect acceleration of two axes.

The IRIS node platform is an IOT node based on the ATmega128l microprocessor chip and the RF230 radio frequency chip. It is a small wireless measurement system specially designed for embedded sensor networks. It is a Mote module that works at 2.4 GHz and supports the IEEE802.15.4 protocol, and is used for low-power IOT.

Several new features added to the IRIS platform improve the node performance as a whole. Its features are as follows: ① Compared with the MICA series products, it has 3 times the range and 2 times the storage space; ② In outdoor testing, the node's line of sight can reach 500 m without an amplifier; ③ RF transmitter based on IEEE802.15.4/ZigBee protocol; ④ 2.4~2.48 GHz. Globally compatible ISM band; ⑤ Direct sequence spread spectrum technology, anti-RF interference, good data shielding; ⑥ 250 kbps data transmission rate; ⑦ Support reliable multi-hop Mesh network; ⑧ Plug and play, can connect sensor boards, data acquisition boards, gateways and software. In addition, IRIS's 51-pin expansion interface can connect analog input, digital I/O, I2C, SPI and UART interfaces, which make it easy to connect with other peripherals. In view of the advantages of the IRIS platform, it is selected as the hardware node of the monitoring system.

5 Linear Discriminant Classification Algorithm

The physical quantities that need to be monitored on-site for transmission lines include local temperature, amplitude and frequency of the line, wind speed, etc. Taking ice warning as an example, according to the different climate and physical environments in different regions, it is necessary to establish expert systems with different parameters based on data. The linear discriminant classification algorithm (LDA) as a multi-source warning decision scheme has the characteristics of simple and efficient algorithm and high confidence.

Discriminant analysis is a commonly used statistical analysis method that determines which category the research object belongs to based on the observed or measured values ​​of several variables. To conduct discriminant analysis, the classification of the observed object and the values ​​of several variables that indicate the characteristics of the observed object must be known. Discriminant analysis is to select variables that can provide more information and establish a discriminant function so that the error rate when using the derived discriminant function to discriminate the category of the observed quantity is minimized.

Suppose there are two types of D-dimensional training samples xk (k=1, 2, ..., n), of which n1 samples are from type wi, n2 samples are from type wi, and n=n1+n2. The two types of training samples constitute the subsets X1 and X2 of the training samples respectively. Let yk be the scalar obtained by transforming vector xk by w, which is one-dimensional. In fact, for a given w, yk is the value of the decision function. The sample mean vector of each type in the D-dimensional feature space is:

After mapping, the larger the distance between the average values ​​of the two categories, the better, and the smaller the intra-class dispersion of the samples of each category, the better. Therefore, the Fisher criterion function is defined as:

The solution w* that maximizes JF is the optimal solution vector, which is Fisher's linear discriminant.

[page]6 Line monitoring solution based on LDA

Wireless communication IOTs are deployed on the transmission lines to collect the temperature of the transmission lines, the amplitude and frequency of the lines, the wind speed, and the line tension. In winter, data is collected and the numerical values ​​of various physical quantities in the two states of deicing and no deicing are stored to establish a training set.

7 Conclusion

A transmission line on-site monitoring and early warning scheme based on the Internet of Things is proposed. Taking advantage of the low power consumption, low cost, multi-sensor, and wireless communication of the Internet of Things, combined with the specific problems currently faced by transmission line monitoring, the selection of the system's hardware platform and the early warning discrimination algorithm are proposed. The monitoring and early warning system scheme can establish a training set based on the specific local environmental characteristics, thereby establishing a highly reliable discriminant function for effective monitoring and early warning.