A5081263995- Power Line Inspection Robots
- Advanced Data and IoT Technologies
- IoT and Edge/Fog Computing
- Image and Object Detection Techniques
- Smart Grid and Power Systems
- Advanced Neural Network Applications
- Electricity Theft Detection Techniques
- Speech Recognition and Synthesis
- Advanced Measurement and Detection Methods
- AI and Big Data Applications
- Privacy-Preserving Technologies in Data
- Full-Duplex Wireless Communications
- Music and Audio Processing
- Anomaly Detection Techniques and Applications
- Industrial Vision Systems and Defect Detection
- Vehicle License Plate Recognition
- Artificial Intelligence in Healthcare
- Advanced Optical Network Technologies
- Software System Performance and Reliability
- Robotic Path Planning Algorithms
- Speech and Audio Processing
- Advanced Technologies in Various Fields
- Cloud Computing and Resource Management
- Image Processing and 3D Reconstruction
- E-commerce and Technology Innovations
China Southern Power Grid (China)
2023-2024
In order to reduce the business platform's processing of massive IoT terminals and meet physical isolation differential quality assurance between different services, 5G slicing technology is used logically isolate network resources effectively protect differentiated service needs users, aiming at problem resource allocation during transmission in power grids. Maximize total throughput EMBB users minimize URLLC by jointly optimizing transmit downlink timeslot base station. Considering that...
Abstract To better analyze and process power data to obtain effective information, a big analysis scheme based on Hadoop architecture is proposed. We the cloud computing environment distributed platform massive of large-scale system. According characteristics intelligent consumption data, common mining algorithm modules such as parallel classification real-time clustering are designed, implementation with different principles further analyzed. Then, HBase adopted access in way, MapReduce...
Abstract To ensure the safe and stable operation of power grid system strengthen refined management level State Grid Corporation China, a big data association analysis model based on Apriori algorithm is proposed. This paper discusses logic framework platform, describes specific implementation from functional architecture, technical architecture other aspects. Aiming at problem massive heterogeneous low sharing degree, transaction matrix pruned compressed by setting minimum support degree...
This article proposes an intelligent data directory construction scheme aimed at supporting the design of ultra-high voltage converter stations. The utilizes visual techniques such as knowledge graphs to analyze large-scale in power industry and establish a classification grading standard. Specifically, selectively identifies key or core for association analysis reduce processing scale improve speed system. After determining results grading, employs machine learning algorithms perform...
The fittings detection of transmission lines plays a vital role in ensuring the safe and stable operation lines. fitting method based on deep learning only scales original image to smaller size. However, it ignores high-definition resolution aerial line, resulting loss rich features high-resolution image. In order solve this problem, we observe that images are concentrated particular area Therefore, propose cascading YOLOx model, including Dense Target Regions (DTR-YOLOx), which can detect...
This paper proposes a new design scheme for an OCR platform. The first pre-trains the model on standard dataset to improve its basic recognition ability, and then fine-tunes pre-trained using transfer learning techniques custom scenario, thus improving ability in power industry applications. approach greatly reduces need annotated data can quickly adapt meter types updates. Additionally, intelligent collection, processing, analysis platform effectively help companies monitor, measure,...
Bolts play an important role in transmission lines, and bolt defects can easily cause abnormality or even failure of lines. However, a large number defect data are difficult to obtain. Aiming at the problem with few samples, this paper proposes method for image generation lines based on different attributes (DBA-GAN). First, images classified according attributes, these categories used as auxiliary information Generative Adversarial Network (GAN). Then, DBA-GAN is generate specified defects....
Substation meter recognition is very significant to status monitoring of instrument equipment in power system. However, meters substation are various, small and dense distribution. Therefore, the process recognizing types based on artificial intelligence, there many problems, such as low pixel, few key features a great deal confusion. How overcome these problems improve accuracy critical. Moreover, Algorithm deployment edge device also trouble. In this paper, lightweight Yolov5-based...
This paper studies the task scheduling strategy of cloud data center based on warehouse. Combined with characteristics power center, a multi-qos evaluation model for is defined. In this paper, neural network algorithm used QoS overall effect value as target. Then establishes technical architecture and hardware software system big electric company, key technologies such access preprocessing, integration storage, computing processing, mining analysis, security management. Finally, taking...
In continuous speech, the speed of speech varies greatly among different speakers in contexts. Deviating from normal often leads to recognition errors and a decrease performance. Considering that impact on segment length units increases or decreases synchronously, there is strong correlation between lengths adjacent units, this paper proposes adaptive algorithm based hidden Markov model distribution, starting utilizing relevant information lengths. Experiments digit strings large vocabulary...