Day-ahead regional solar power forecasting with hierarchical temporal convolutional neural networks using historical power generation and weather data

TL;DR

This work proposes two deep-learning-based regional forecasting methods that can effectively leverage both types of time series (aggregated and individual) with weather data in a region and compares their approaches with well-known alternative methods.

Abstract

Regional solar power forecasting, which involves predicting the total power generation from all rooftop photovoltaic systems in a region holds significant importance for various stakeholders in the energy sector. However, the vast amount of solar power generation and weather time series from geographically dispersed locations that need to be considered in the forecasting process makes accurate regional forecasting challenging. Therefore, previous work has limited the focus to either forecasting a single time series (i.e., aggregated time series) which is the addition of all solar generation time series in a region, disregarding the location-specific weather effects or forecasting solar generation time series of each PV site (i.e., individual time series) independently using location-specific weather data, resulting in a large number of forecasting models. In this work, we propose two deep-learning-based regional forecasting methods that can effectively leverage both types of time series (aggregated and individual) with weather data in a region. We propose two hierarchical temporal convolutional neural network architectures (HTCNN) and two strategies to adapt HTCNNs for regional solar power forecasting. At first, we explore generating a regional forecast using a single HTCNN. Next, we divide the region into multiple sub-regions based on weather information and train separate HTCNNs for each sub-region; the forecasts of each sub-region are then added to generate a regional forecast. The proposed work is evaluated using a large dataset collected over a year from 101 locations across Western Australia to provide a day ahead forecast. We compare our approaches with well-known alternative methods and show that the sub-region HTCNN requires fewer individual networks and achieves a forecast skill score of 40.2% reducing a statistically significant error by 6.5% compared to the best counterpart.

Figures (17)

• Figure 1: The hierarchical structure of solar power generation time series datasets in a region.
• Figure 1: Standard 1D causal convolution and dilated 1D causal convolution with a dilation rate of 2 on an input sequence $x_0, x_1, x_2,\dots,x_{t-1},x_t$. The output feature map generated after applying the filter is shown in $y$.
• Figure 2: An example of normalised regional solar power generation and solar power generation at three different postcodes of this region.
• Figure 3: A traditional 1D convolutional neural network architecture for forecasting which consists of standard 1D convolutions and pooling layers. In addition, other layers such as dropout, normalization maybe included to prevent over fitting of the network.
• Figure 4: Visualisation of a stack of dilated causal convolutions (with dilation rates 1, 2, 4 and padding where zero padding is shown with dash squares) on an input sequence $x_0, x_1, x_2,\dots,x_{t-1},x_t$. The blue lines show a filter with a filter size of 2. The output sequences generated after applying the filters are shown as $y$/ $y'$/ $y"$.