Fuzzy Clustering-Based Deep Learning for Short-Term Load Forecasting in Power Grid Systems Using Time-Varying and Time-Invariant Features.

Accurate short-term load forecasting (STLF) is essential for power grid systems to ensure reliability, security and cost efficiency. Thanks to advanced smart sensor technologies, time-series data related to power load can be captured for STLF. Recent research shows that deep neural networks (DNNs) are capable of achieving accurate STLP since they are effective in predicting nonlinear and complicated time-series data. To perform STLP, existing DNNs use time-varying dynamics of either past load consumption or past power correlated features such as weather, meteorology or date. However, the existing DNN approaches do not use the time-invariant features of users, such as building spaces, ages, isolation material, number of building floors or building purposes, to enhance STLF. In fact, those time-invariant features are correlated to user load consumption. Integrating time-invariant features enhances STLF. In this paper, a fuzzy clustering-based DNN is proposed by using both time-varying and time-invariant features to perform STLF. The fuzzy clustering first groups users with similar time-invariant behaviours. DNN models are then developed using past time-varying features. Since the time-invariant features have already been learned by the fuzzy clustering, the DNN model does not need to learn the time-invariant features; therefore, a simpler DNN model can be generated. In addition, the DNN model only learns the time-varying features of users in the same cluster; a more effective learning can be performed by the DNN and more accurate predictions can be achieved. The performance of the proposed fuzzy clustering-based DNN is evaluated by performing STLF, where both time-varying features and time-invariant features are included. Experimental results show that the proposed fuzzy clustering-based DNN outperforms the commonly used long short-term memory networks and convolution neural networks.

Study of Current Measurement Method Based on Circular Magnetic Field Sensing Array.

Classic core-based instrument transformers are more prone to magnetic saturation. This affects the measurement accuracy of such transformers and limits their applications in measuring large direct current (DC). Moreover, protection and control systems may exhibit malfunctions due to such measurement errors. This paper presents a more accurate method for current measurement based on a circular magnetic field sensing array. The proposed measurement approach utilizes multiple hall sensors that are evenly distributed on a circle. The average value of all hall sensors is regarded as the final measurement. The calculation model is established in the case of magnetic field interference of the parallel wire, and the simulation results show that the error decreases significantly when the number of hall sensors n is greater than 8. The measurement error is less than 0.06% when the wire spacing is greater than 2.5 times the radius of the sensor array. A simulation study on the off-center primary conductor is conducted, and a kind of hall sensor compensation method is adopted to improve the accuracy. The simulation and test results indicate that the measurement error of the system is less than 0.1%.

Machine learning based multi-method interpretation to enhance dissolved gas analysis for power transformer fault diagnosis.

Accurate interpretation of dissolved gas analysis (DGA) measurements for power transformers is essential to ensure overall power system reliability. Various DGA interpretation techniques have been proposed in the literature, including the Doernenburg Ratio Method (DRM), Roger Ratio Method (RRM), IEC Ratio Method (IRM), Duval Triangle Method (DTM), and Duval Pentagon Method (DPM). While these techniques are well documented and widely used by industry, they may lead to different conclusions for the same oil sample. Additionally, the ratio-based methods may result in an out-of-code condition if any of the used gases fall outside the specified limits. Incorrect interpretation of DGA measurements can lead to mismanagement and may lead to catastrophic consequences for operating power transformers. This paper presents a new interpretation technique for DGA aimed at improving its accuracy and consistency. The proposed multi-method approach employs s scoring index and random forest machine learning principles to integrate existing interpretation methods into one comprehensive technique. The robustness of the proposed method is assessed using DGA data collected from several transformers under various health conditions. Results indicate that the proposed multi-method, based on the scoring index and random forest; offers greater accuracy and consistency than individual conventional interpretation methods alone. Furthermore, the multi-method based on random forest demonstrated higher accuracy than employing the scoring index only.