Search

One key sequence control of substation is the latest requirement of intelligent substation. The double confirmation in its operation specification requires that when the equipment is operated remotely, there should be at least two non homologous indicators to confirm that the equipment has been operated accurately. In this paper, a fish eye switch on / off status recognition algorithm based on image intelligent analysis is proposed. This algorithm realizes the judgment of on and off as well as abnormal and occlusion status. The on and off angle can also be quantified. The accuracy rate is 99.99% after field verification.

To solve the weather classification problem in outdoor power monitoring, this paper proposes a weather classification algorithm based on improved SqueezeNet. In the proposed network, three modifications are made: Firstly, the input size is increased in the first convolution layer and the convolution kernel is reduced to make it more suitable for high-resolution image classification. Secondly, the combination of global average pooling and small fully connected layers leads to a proper tradeoff between computational burden and classification performance of the proposed network. Thirdly, the introduction of batch normalization not only suppresses the over-fitting phenomenon, but also increases classification accuracy and converging speed. According to the actual application scenario, the multi-weather image dataset is constructed and used for training and test. Experimental results verify the effectiveness of the proposed network, and reveal the proposed network, compared with the original SqueezeNet, could improve the performance of classification accuracy and suppress the over-fitting.

There are still a large number of pointer meters in substations, which mainly rely on manual inspection to detect the readings of pointer meters. In order to solve the above problems, this paper proposes a pointer meter keypoints detection method based on deep learning combined with virtual pointer rotation to determine the meter readings. This method first uses the deep learning algorithm to train data set of pointer meters and obtain the keypoints information of pointer meters. According to the keypoints information, we can further determine the center of rotation and radius of the pointer meter, then the real position of the pointer meter can be determined. Finally, according to the meter range information, as well as the real pointer angle and the meter range angle, we can calculate the readings of the pointer meter. The experiment results show that the proposed method can quickly and accurately obtain the pointer position, and further identify the meter readings. The proposed method can reduce the uncertainty caused by manual inspection and have a great impact on intelligent development of substations.

With the rapid development of science and technology, image processing technology plays an important role in the field of computer vision. In order to improve the matching speed and real-time requirements, this paper proposes an image matching algorithm based on ORB and K-means clustering, which can effectively improve the accuracy of image feature point location and the accuracy and efficiency of image feature matching, and reduce the time consumption. The algorithm uses sub-pixel interpolation to optimize the traditional ORB algorithm, which improves the accuracy and characteristics of clustering calculation.

Power transmission line safety monitoring is one of the important tasks to maintain the security of national power grid. In this paper, the object detection method based on computer vision is applied to automatically monitor the potential safety risk of transmission line. We firstly create a potential safety risk object dataset. Secondly we analyze most state-of-the-art object detection model. Thirdly according to the specific dataset, an object detection model was trained, which uses training tricks to get high performance. Fourthly, we built a monitoring system that feeds the discriminant results back to the display terminal, which can comprehensively grasp the situation of the whole safe area and ensure the safe operation of the transmission network. Our experiments show the excellent results are Cascade R-CNN detection framework based on deep learning and backbone based on high resolution representations network. It gains 81.5 mAP on 26 kinds of objects datasets at IOU threshold 0.5, and show hidden danger detection algorithm based on deep learning can accurately discriminate the dangerous sources. The monitoring system feeds the discriminant results back to the display terminal, which can comprehensively grasp the situation of the whole safe area and ensure the safe operation of the transmission network.