Infrared camera can be used to monitor the condition of substation electrical equipment. Fast and accurate target detection algorithm is the key for infrared intelligent on-line routing inspection. However, the performance of traditional detection algorithm is poor due to the complex background of substation. To solve this problem, this paper proposes a target detection model based on the improved faster region-based convolutional neural networks (Faster R-CNN), which can be used for automatic detection of five kinds of electrical equipment in substations. The feature extraction network of this model is improved based on VGG16 by abandoning some high-level convolutions to accelerate the training and testing speed. Meanwhile, the 1:3 and 3:1 aspect ratio of anchor are added to improve the detection accuracy of elongated equipment. Experiments are performed on an infrared image dataset of substation to detect five types of equipment. The robustness tests of our model are carried out, too. The results show that our model performs well in detection accuracy and speed, achieving a mean detected accuracy of 95.32% and running at 11 frame/second. In addition, our model is robust to noise and lightness, which is suitable to other substations. Comparing to other models, our model has the highest mAP@0.5 of 92.78%.

Online fault detection of insulators is a necessary requirement for the development of a smart grid, which directly affects the safety and reliability of power system operations. Breakage is an important factor causing insulator abnormalities, and a large number of image samples are often required for training to complete defect identification with the help of AI deep learning. In this paper, we propose a convolutional neural network ensemble implemented using small samples, data augmentation and transfer learning for insulator breakage detection. We used data augmentation to expand the small sample of 200 insulator images by 15 times to 3000 images, built a VGG16 transfer network, pre-trained the VGG16 network convolutional basis weights, and improved the training performance and efficiency of the model with freeze training and fine-tuning training. The training on 2700 images and testing results on 300 images show that the proposed method can achieve an accuracy of 0.9871 for insulator breakage recognition.

Insulators are one of the widely used important devices in power systems, and proposing an accurate and effective online monitoring method for insulator defects is an urgent need for intelligent and informative power systems. We propose a multiscale residual neural network for insulator surface breakage recognition. The proposed multiscale convolution filter uses three different size convolution kernels for convolution filtering, and performs feature maps fusion, which can enrich the spatial correlation and channel correlation of feature maps. We investigated the effects of different learning rates on the model performance, and confirmed that the model performs best with learning rate of 10-6. Compared with VGG16, Inception_v3 and Resnet50, the multiscale residual neural network model has significant advantages in detection capability and occupies less computational resources, and the detection speed can meet the industrial real-time detection requirements. Through dark-light and noise experiments, it is demonstrated that the proposed method is robust and can effectively deal with low-quality insulator images.

This paper presents the observation results of the induced voltage at the terminal of a 10 kV distribution line with 40 m distance from nearby triggered lightning pole at Guangzhou in the Summer of 2018. It is shown that the induced voltage waveforms exhibit three components including a slow increase preceding the initial negative peak, following a rapid polarity reversal and attenuated oscillations. The characteristic waveform parameters are defined in this paper. The observed negative peak value voltage ranges from -47.1 kV to -7.5 kV. The positive peak value voltage ranges from 5.3 kV to 20.9 kV, and the peak to peak value voltage ranges from 12.8 kV to 63.2 kV. The test results show that the negative peak value voltage of the induced voltage at the terminal of distribution generated by nearby negative triggered lightning generally exceeds the positive peak value, and in some cases the negative peak value even reaches twice the positive peak value. The negative 10%-90% peak rise time ranges from 2.3 's to 88.5 μs, and the average oscillation period is about 18 μs for oscillation component. Both negative and positive peaks of induced voltage show good linear relationship with the lightning current.

In the summer of 2013, a three-dimensional (3D)-based Foshan Total Lightning Location System (FTLLS), embedded with differential time of arrival (DTOA) techniques, was installed and started its operation in Foshan, Guangdong Province, China. In this paper, the geographical distribution and set-up information of FTLLS, the estimated locating errors and locating results, as well as its initial operation results are presented. FTLLS consists of nine sub-stations that receive electromagnetic waves associated with lightning discharges and locates VLF/LF (200 Hz–500 kHz) radiation sources in 3D. The remote sub-stations acquired triggered waveforms with a duration of 0.5 ms, a resolution of 12-bits, and a GPS-based sferic time tags of 24 h per day. Cloud-to-ground (CG) lightning events, intra-cloud (IC) lightning events and narrow bipolar events (NBEs) were located by FTLLS. Based on the Monte Carlo simulation, the two-dimensional horizontal location error is basically less than 100 m, and the vertical error (altitude) is less than 200 m when the lightning event occurs within the network. On the other hand, over 14 million lightning strikes were recorded successfully by FTLLS during the period of May to October in 2014, among which IC events, CG events and NBEs accounted for 65%, 34% and 1%, respectively. It is shown that FTLLS is capable of a fine three-dimensional (3D) location, in which the altitude parameters obtained are reasonable and well consistent with observed data in the previous studies. The location results of thunderstorms were additionally verified through simultaneously-observed radar data.

In recent years, frequent accidents due to lightning have occurred in wind power plants, resulting in damages to electrical equipments in the power collection system, especially in the cable power collection system (CPCS) and the nacelle step‐up transformer. In this study, electromagnetic transient simulation of lightning overvoltage in a 35 kV CPCS with the nacelle step‐up transformer was carried out. Influencing factors including the location of the step‐up transformer, the lightning current, the grounding resistance of wind turbine generator (WTG) and distance between WTGs are taken into consideration. The simulation results show that, in the case of 30 Ω grounding resistance for the WTG, the current flowing through the arrester installed at the high voltage winding side of the step‐up transformer will reach up to 27.1 kA when subjected to a 200 kA lightning current, which reveals the damage mechanism of arrester blasting. Based on the calculation results, recommended nominal discharge current value for the arrester as well as the optimal lightning overvoltage protection method are proposed here.