Abstract:Accurate photovoltaic (PV) power forecasting is critical for integrating renewable energy sources into the grid, optimizing real-time energy management, and ensuring energy reliability amidst increasing demand. However, existing models often struggle with effectively capturing the complex relationships between target variables and covariates, as well as the interactions between temporal dynamics and multivariate data, leading to suboptimal forecasting accuracy. To address these challenges, we propose a novel model architecture that leverages the iTransformer for feature extraction from target variables and employs long short-term memory (LSTM) to extract features from covariates. A cross-attention mechanism is integrated to fuse the outputs of both models, followed by a Kolmogorov-Arnold network (KAN) mapping for enhanced representation. The effectiveness of the proposed model is validated using publicly available datasets from Australia, with experiments conducted across four seasons. Results demonstrate that the proposed model effectively capture seasonal variations in PV power generation and improve forecasting accuracy.

What problem does this paper attempt to address?

This paper attempts to solve the problem of complex relationship modeling in photovoltaic (PV) power prediction. Specifically, existing models are deficient in effectively capturing the complex relationships between target variables and covariates as well as the interactions between time dynamics and multivariate data, resulting in lower prediction accuracy. To overcome these challenges, the authors propose a new model architecture that combines iTransformer and Long - Short - Term Memory (LSTM), and fuses the outputs of the two through a Cross - Attention Mechanism. Finally, it is enhanced by the Kolmogorov - Arnold Network (KAN) for enhanced representation to improve prediction accuracy. ### Main problems 1. **Complex relationship modeling**: Existing models have difficulty in effectively capturing the complex relationships between target variables and covariates. 2. **Interaction between time dynamics and multivariate data**: Existing models perform poorly in dealing with the interaction between time dynamics and multivariate data. 3. **Prediction accuracy**: Due to the above problems, the prediction accuracy of existing models is relatively low, especially in cases where there are large seasonal changes and fluctuations in weather conditions. ### Solutions - **iTransformer**: Used to extract features from target variables, emphasizing the capture of multivariate correlations. - **LSTM**: Used to extract features from covariates, especially suitable for dealing with long - term dependencies in time - series data. - **Cross - attention mechanism**: Fuses the outputs of iTransformer and LSTM, enhancing the ability to model time dynamics and multivariate relationships. - **Kolmogorov - Arnold network**: Further maps the fused features to the required output dimensions, improving the representational ability of the model. ### Experimental verification The authors conducted experiments using publicly available datasets in Australia, covering data from four seasons. The results show that the proposed model can effectively capture seasonal changes and significantly improve prediction accuracy. ### Main contributions 1. **Effective multivariate feature extraction**: By combining iTransformer and LSTM, the model performs well in feature extraction of target variables and covariates. 2. **Powerful interaction modeling**: The cross - attention mechanism enables the model to effectively capture the interaction between time dynamics and multivariate relationships. 3. **Effectiveness verification and seasonal analysis**: Through experiments in four seasons, the effectiveness and robustness of the model are verified. In conclusion, this paper aims to solve the problems of complex relationship modeling and time - dynamic interaction in photovoltaic power prediction through an innovative model architecture, thereby improving prediction accuracy.

Enhanced Photovoltaic Power Forecasting: An iTransformer and LSTM-Based Model Integrating Temporal and Covariate Interactions

A hybrid deep learning model with an optimal strategy based on improved VMD and transformer for short-term photovoltaic power forecasting

Cross-variable Linear Integrated ENhanced Transformer for Photovoltaic power forecasting

A Novel Hybrid Transformer-Based Framework for Solar Irradiance Forecasting under Incomplete Data Scenarios

A Novel Improved Variational Mode Decomposition-Temporal Convolutional Network-Gated Recurrent Unit with Multi-Head Attention Mechanism for Enhanced Photovoltaic Power Forecasting

TransPVP: A Transformer-Based Method for Ultra-Short-Term Photovoltaic Power Forecasting

Contextual feature fusion convolutional transformer complementation for PV power prediction

A hybrid machine learning forecasting model for photovoltaic power

An improved temporal convolutional network with attention mechanism for photovoltaic generation forecasting

Short-Term Photovoltaic Power Generation Prediction Based on Copula Function and CNN-CosAttention-Transformer

Photovoltaic power forecasting: A hybrid deep learning model incorporating transfer learning strategy

Solar Irradiance Prediction with Multi-Meteorological Variables Based on iTransformer

Intra-hour solar irradiance forecasting: An end-to-end Transformer-based network

Interpretable Feature-Temporal Transformer for Short-Term Wind Power Forecasting with Multivariate Time Series

Photovoltaic Power Prediction Based on Irradiation Interval Distribution and Transformer-LSTM

An Attention-Based Photovoltaic Forecasting Scheme Combined with LSTM Model

Enhancing One-Day-Ahead Probabilistic Solar Power Forecast With a Hybrid Transformer-LUBE Model and Missing Data Imputation

A Point-interval Prediction Method for Photovoltaic Power Generation Considering Weather Types: a Novel Hybrid Deep Learning Model

Photovoltaic Power Forecasting Based on IEWT-Attention-LSTM Model

Multistep Photovoltaic Power Forecasting Based on Multi-Timescale Fluctuation Aggregation Attention Mechanism and Contrastive Learning

Efficient Subseasonal Weather Forecast using Teleconnection-informed Transformers