Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MATNet: Multi-Level Fusion Transformer-Based Model for Day-Ahead PV Generation Forecasting

Matteo Tortora, Francesco Conte, Gianluca Natrella, Paolo Soda

TL;DR

This paper proposes MATNet, a novel self-attention transformer-based architecture for multivariate multi-step day-ahead PV power generation forecasting that combines the AI paradigm with the prior physical knowledge of PV power generation of physics-based methods.

Abstract

Accurate forecasting of renewable generation is crucial to facilitate the integration of RES into the power system. Focusing on PV units, forecasting methods can be divided into two main categories: physics-based and data-based strategies, with AI-based models providing state-of-the-art performance. However, while these AI-based models can capture complex patterns and relationships in the data, they ignore the underlying physical prior knowledge of the phenomenon. Therefore, in this paper we propose MATNet, a novel self-attention transformer-based architecture for multivariate multi-step day-ahead PV power generation forecasting. It consists of a hybrid approach that combines the AI paradigm with the prior physical knowledge of PV power generation of physics-based methods. The model is fed with historical PV data and historical and forecast weather data through a multi-level joint fusion approach. The effectiveness of the proposed model is evaluated using the Ausgrid benchmark dataset with different regression performance metrics. The results show that our proposed architecture significantly outperforms the current state-of-the-art methods. These findings demonstrate the potential of MATNet in improving forecasting accuracy and suggest that it could be a promising solution to facilitate the integration of PV energy into the power grid.

MATNet: Multi-Level Fusion Transformer-Based Model for Day-Ahead PV Generation Forecasting

TL;DR

This paper proposes MATNet, a novel self-attention transformer-based architecture for multivariate multi-step day-ahead PV power generation forecasting that combines the AI paradigm with the prior physical knowledge of PV power generation of physics-based methods.

Abstract

Accurate forecasting of renewable generation is crucial to facilitate the integration of RES into the power system. Focusing on PV units, forecasting methods can be divided into two main categories: physics-based and data-based strategies, with AI-based models providing state-of-the-art performance. However, while these AI-based models can capture complex patterns and relationships in the data, they ignore the underlying physical prior knowledge of the phenomenon. Therefore, in this paper we propose MATNet, a novel self-attention transformer-based architecture for multivariate multi-step day-ahead PV power generation forecasting. It consists of a hybrid approach that combines the AI paradigm with the prior physical knowledge of PV power generation of physics-based methods. The model is fed with historical PV data and historical and forecast weather data through a multi-level joint fusion approach. The effectiveness of the proposed model is evaluated using the Ausgrid benchmark dataset with different regression performance metrics. The results show that our proposed architecture significantly outperforms the current state-of-the-art methods. These findings demonstrate the potential of MATNet in improving forecasting accuracy and suggest that it could be a promising solution to facilitate the integration of PV energy into the power grid.
Paper Structure (19 sections, 7 equations, 4 figures, 3 tables)

This paper contains 19 sections, 7 equations, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Materials
  4. Ausgrid
  5. Weather Conditions
  6. Proposed Architecture
  7. Embedding Module
  8. Positional Encoding
  9. Self-Attention
  10. Dense Interpolation Layer
  11. Multi-Level Joint Fusion
  12. Experimental setup and results
  13. Training Details
  14. Evaluation Metrics
  15. Ausgrid Benchmark
  16. ...and 4 more sections

Figures (4)

  • Figure 1: Map of the territorial areas containing the 26 pv units considered in this study, generated using the PyTrack library tortora2022pytrack. Each coloured area corresponds to a distinct geographical region associated with one of the eight zip codes.
  • Figure 2: Overall architecture of our proposed MATNet.
  • Figure 3: Visualising the dense interpolation module. Where $m$, $n$, $d$, are the length of the time series, the interpolation factor, and the number of time series attributes, respectively. In this work, we have that $m=step_{in}$, $d=d_{model}$ and $n=M$.
  • Figure 4: Forecasting results of MATNet on Ausgrid test set. On the left is the forecast for the best-performing day (2013-06-11), while on the right is the forecast for the worst-performing day (2013-01-28). Both predictions are evaluated using the MASE metric.