Sort by
Toward a Comprehensive Evaluation on the Online Methods for Monitoring Transformer Turn-to-Turn Faults

Transformer winding turn-to-turn fault is the prominent cause of transformer total failure, so detecting the winding fault in real time to stop the failure development in advance is imperative. However, existing techniques entailing periodic offline inspections fail to continuously monitor transformer winding states while causing extra costs due to the outage during inspections. This has driven researchers to consider effective continuous online monitoring methods from several technical perspectives, including typically port voltage current analysis, online frequency response analysis, and vibration analysis. Since these methods are conventionally evaluated with qualitative comparisons focusing only on feasibility, quantitative assessments indispensable for the targeted improvement of the methods and the most suitable method decision in specific scenarios are still missing. To this end, we conduct a comprehensive evaluation on the three methods by leveraging both experiment and theoretical analysis. Specifically, a customized experiment platform has been designed to support data acquisition under different operating conditions. As conventional feature mining algorithms cannot process the monitoring data produced by different methods in a uniform manner, a feature extraction algorithm leveraging image mining is proposed to extract data features after mapping the test data into a high-dimensional image. This novel algorithm allows us to fully assess several fundamental aspects (i.e., sensitivity, repeatability, and anti-interference capability) of these monitoring methods.

Relevant
A voltage control method for distribution networks based on TCN and MPGA under cloud edge collaborative architecture

A distribution network voltage control method based on TCN and MPGA under cloud edge collaborative architecture is proposed to address the issues of heavy computational burden and inability to balance economic efficiency in traditional decentralized and centralized voltage control methods. Firstly, under the cloud edge collaboration architecture, the power prediction model based on time convolutional networks is trained in the cloud, and the edge predicts the load demand and new energy active output in each region based on the TCN Attention power prediction model issued by the cloud. Then, considering safety and economy comprehensively, with the goal of minimizing the daily network loss and voltage regulation cost in the distribution network, the Pareto solution set for the optimal operation point of each voltage regulation equipment in the entire distribution network was solved based on an improved multi population genetic algorithm at the distribution cloud main station. Finally, the optimal distribution network voltage control scheme is obtained based on the Pareto solution set. Based on the simulation system, experimental testing was conducted on the proposed method, and the results showed that its daily voltage regulation cost and network loss were 863.4 yuan and 43.73 MW h, respectively. The average absolute percentage error of the prediction model was 4.51 %, which was superior to other comparative methods.

Relevant
Multi-Type Missing Imputation of Time-Series Power Equipment Monitoring Data Based on Moving Average Filter-Asymmetric Denoising Autoencoder.

Supervisory control and data acquisition (SCADA) systems are widely utilized in power equipment for condition monitoring. For the collected data, there generally exists a problem-missing data of different types and patterns. This leads to the poor quality and utilization difficulties of the collected data. To address this problem, this paper customizes methodology that combines an asymmetric denoising autoencoder (ADAE) and moving average filter (MAF) to perform accurate missing data imputation. First, convolution and gated recurrent unit (GRU) are applied to the encoder of the ADAE, while the decoder still utilizes the fully connected layers to form an asymmetric network structure. The ADAE extracts the local periodic and temporal features from monitoring data and then decodes the features to realize the imputation of the multi-type missing. On this basis, according to the continuity of power data in the time domain, the MAF is utilized to fuse the prior knowledge of the neighborhood of missing data to secondarily optimize the imputed data. Case studies reveal that the developed method achieves greater accuracy compared to existing models. This paper adopts experiments under different scenarios to justify that the MAF-ADAE method applies to actual power equipment monitoring data imputation.

Open Access
Relevant