# An Explainable, Attention-Enhanced, Bidirectional Long Short-Term Memory Neural Network for Joint 48-Hour Forecasting of Temperature, Irradiance, and Relative Humidity

## TL;DR

This paper introduces an attention-enhanced BiLSTM neural network for accurate 48-hour joint forecasting of temperature, irradiance, and humidity, supporting smart HVAC control with improved interpretability.

## Contribution

It proposes a novel multivariate deep learning framework combining attention mechanisms and BiLSTM for joint weather forecasting, with enhanced explainability and superior accuracy.

## Key findings

- Achieved MAEs of 1.3°C, 31 W/m², and 6.7% for temperature, irradiance, and humidity.
- Outperformed existing numerical weather prediction and machine learning benchmarks.
- Provided interpretability through feature attribution and attention analysis.

## Abstract

This paper presents a Deep Learning (DL) framework for 48-hour forecasting of temperature, solar irradiance, and relative humidity to support Model Predictive Control (MPC) in smart HVAC systems. The approach employs a stacked Bidirectional Long Short-Term Memory (BiLSTM) network with attention, capturing temporal and cross-feature dependencies by jointly predicting all three variables. Historical meteorological data (2019-2022) with encoded cyclical time features were used for training, while 2023 data evaluated generalization. The model achieved Mean Absolute Errors of 1.3 degrees Celsius (temperature), 31 W/m2 (irradiance), and 6.7 percentage points (humidity), outperforming state-of-the-art numerical weather prediction and machine learning benchmarks. Integrated Gradients quantified feature contributions, and attention weights revealed temporal patterns, enhancing interpretability. By combining multivariate forecasting, attention-based DL, and explainability, this work advances data-driven weather prediction. The demonstrated accuracy and transparency highlight the framework's potential for energy-efficient building control through reliable short-term meteorological forecasting.

---
Source: https://tomesphere.com/paper/2508.21109