Islanding Detection for Active Distribution Networks Using WaveNet+UNet Classifier

TL;DR

A WaveNet classifier reinforced by a denoising U-Net model is proposed to address the gaps in the field to ensure reliable islanding detection in active distribution networks and is robust against noisy conditions by incorporating a denoising U-Net model.

Abstract

This paper proposes an AI-based scheme for islanding detection in active distribution networks. By reviewing existing studies, it is clear that there are several gaps in the field to ensure reliable islanding detection, including (i) model complexity and stability concerns, (ii) limited accuracy under noisy conditions, and (iii) limited applicability to systems with different types of resources. Accordingly, this paper proposes a WaveNet classifier reinforced by a denoising U-Net model to address these shortcomings. The proposed scheme has a simple structure due to the use of 1D convolutional layers and incorporates residual connections that significantly enhance the model's generalization. Additionally, the proposed scheme is robust against noisy conditions by incorporating a denoising U-Net model. Furthermore, the model is sufficiently fast using a sliding window time series of 10 milliseconds for detection. Utilizing positive/negative/zero sequence components of voltages, superimposed waveforms, and the rate of change of frequency provides the necessary features to precisely detect the islanding condition. In order to assess the effectiveness of the suggested scheme, over 3k islanding/non-islanding cases were tested, considering different load active/reactive powers values, load switching transients, capacitor bank switching, fault conditions in the main grid, different load quality factors, signal-to-noise ratio levels, and both types of conventional and inverter-based sources.

Figures (9)

• Figure 1: Schematic of the test system.
• Figure 2: Block diagram of the proposed algorithm for islanding detection method: (1)- measuring three-phase voltages, (2)- feature calculation, (3)- Unet + WaveNet classifier, and (4)- islanding status signal output.
• Figure 3: Schematic of (a) the recurrent neural network and (b) an LSTM cell diagram at time step t.
• Figure 4: An example of the stack of causal convolutional layers inspired from vanwavenet. The depicted bold red arrows distinctly delineate the connections orchestrating the relationship between the output at time $t$ and the preceding layers. Notably, this configuration enables the acquisition of the conditional term $p(x_t|x_{1:t-1})$ based on the most recent 16 input time steps, corresponding to a receptive field of size 16.
• Figure 5: Schematic of (a) the gated activation units and residual connections used at each layer of the WaveNet, (b) the proposed WaveNet classifier for Islanding detection. In this model, the FC refers to a fully connected layer.