Insulator faults detection is an important task for high-voltage transmission line inspection. However, current methods often suffer from the lack of accuracy and robustness. Moreover, these methods can only detect one fault in the insulator string, but cannot detect a multi-fault. In this paper, a novel method is proposed for insulator one fault and multi-fault detection in UAV-based aerial images, the backgrounds of which usually contain much complex interference. The shapes of the insulators also vary obviously due to the changes in filming angle and distance. To reduce the impact of complex interference on insulator faults detection, we make full use of the deep neural network to distinguish between insulators and background interference. First of all, plenty of insulator aerial images with manually labelled ground-truth are collected to construct a standard insulator detection dataset ‘InST_detection’. Secondly, a new convolutional network is proposed to obtain accurate insulator string positions in the aerial image. Finally, a novel fault detection method is proposed that can detect both insulator one fault and multi-fault in aerial images. Experimental results on a large number of aerial images show that our proposed method is more effective and efficient than the state-of-the-art insulator fault detection methods.

Accurate and timely detection of insulator flashover on power transmission lines is of paramount importance to power utilities. Most available solutions mainly focus on the exploitation of the flashover mechanism or the discharge area detection, rather than the identification of a damaged area due to flashovers using captured aerial images. To this end, this paper proposes a multi-saliency aggregation-based porcelain insulator flashover fault detection approach. The target area of the insulator is determined using the Faster-Pixelwise Image Saliency by Aggregating (F-PISA) algorithm based on the color and structural features. The color model can be established based on the color feature of the damaged areas on the insulator surface, and hence the damaged area can be identified. Based on the information obtained above, the contour information can be extracted. With the preceding process, the fault location can be confirmed with a good accuracy. The performance of the proposed detection approach is assessed through a comparative study with other available solutions. The numerical result demonstrates that the suggested solution can detect the insulator flashover with improved performance in terms of the average detection rate and average efficient detection rate. Additional analysis is carried out to evaluate its robustness and real-time performance, which confirms its deployment feasibility in practice.

During an automatic power transmission line inspection, a large number of images are collected by unmanned aerial vehicles (UAVs) to detect existing defects in transmission line components, especially insulators. However, with twin insulator strings in the inspection images, when the umbrella skirts of the rear string are obstructed by the front string, defect detection becomes difficult. To solve this problem, we propose a method to detect self-shattering defects of insulators based on spatial features contained in images. Firstly, the images are segmented according to the particular color features of glass insulators, and the main axes of insulator strings in the images are adjusted to the horizontal direction. Then, the connected regions of insulators in the images are marked. After that, the vertical lengths of the regions, the number of insulator pixels in the regions, as well as the horizontal distances between two adjacent connected regions are selected as spatial features, based on which defect discriminants are formulated. Finally, experiments are performed using the proposed formula to detect self-shattering defects in the insulators, using the spatial distribution of the connected regions to locate the defects. The experiment results indicate that the proposed method has good detection accuracy and localization precision.

To address the issue of navigation integrity loss caused by navigation sensor failures during the actual operation, a multi-sensor fusion navigation integrity monitoring method based on GAF-PCNN is proposed. First, according to the subsystem combination in the multi-sensor fusion navigation system, principal component analysis is used to extract the fault principal component features from the KF-LS form solution and establish fault detection statistics, forming one-dimensional time series feature data. Second, the Gramian Angular Field (GAF) is used to convert the one-dimensional time series feature data, generating two sets of image data containing state features and time information. These two sets of image data are then fed into a dual-channel convolutional neural network for parallel training. In a real-time system, the trained model is used for fault detection and isolation. Finally, the system protection level is calculated, and the integrity assessment and warning of the multi-sensor fusion navigation system are realized. In the marine environment, simulation tests were conducted using real collected ship trajectory data. The test results show that the method can promptly and effectively detect and isolate faults. The fault classification accuracy on test set is 96.9%, and the detection time for constant bias faults is around 1s, and the detection time for gradual faults is around 3.5s. The calculate protection level can issue integrity warning prompts within 5s when the positioning error exceed the alarm threshold of 5m, ensuring the robustness and reliability of the multi-sensor fusion navigation system.

The airborne sensor data provides the navigation information required for warning computation to the terrain awareness warning algorithm through navigation fusion. Therefore, when the sensor data is abnormal, it will affect the reliability of the terrain awareness warning. To improve the warning reliability and prevent the controlled flight into terrain accident, a multi sensor fusion navigation integrity monitoring for terrain awareness prediction warning algorithm was proposed. Firstly, the algorithm constructs a multi sensor fusion navigation integrity monitoring method to detect abnormal sensor information, provides reliable navigation information for the warning algorithm according to the abnormal identification results, and ensures the reliability of terrain awareness. Secondly, when the sensor data is integrity, the algorithm uses the BP neural network to predict the navigation information in short time to achieve the terrain awareness prediction warning, increasing the reaction time. The algorithm is tested by using real terrain elevation map and simulated flight trajectory. The test results show that the multi sensor fusion navigation integrity monitoring can effectively detect sensor anomalies within 6 seconds, and provide reliable navigation information for warning calculation, which ensures the reliability of the terrain awareness warning. The adopted prediction warning method can increase the reaction time by about 5 seconds, and the accuracy of the prediction warning is above 90%, improving the pilot’s ability to actively avoid threatening terrain.

The microgrid central controller (MGCC) integrates the functions of control, monitoring, and communication in microgrid system, and has powerful capabilities of information collection and data processing. However, with the development of microgrid system worldwide, the information security management capabilities of the MGCC are poor, If information / network attacks cannot be actively detected and identified, it will easily reduce the reliability of the microgird system operation. Attackers can use abnormal information or use the MGCC as a springboard to further attack the upper-layer system. Aiming at the above problems, this paper presents an attack detection method based on convolutional neural network, and a detailed design process of attack detection model of the MGCC is proposed. In the attack detection method, the important data streams in the MGCC are used as the input of the convolutional neural network model, then the convolutional neural network model detects or classifies these data streams, finally, intercept the data flow with attack behavior and give a warning prompt, and forward data without attack behaviors normally.