Curated Solar Weather and PV AC Power Output Dataset for Explainable Solar Power Forecasting

This dataset contains a curated, processed solar-weather and photovoltaic AC power output dataset prepared for machine-learning-based solar power forecasting. The dataset was developed by integrating publicly available PV generation data and solar-weather data originally obtained from Kaggle. The PV generation component was derived from Plant_1_Generation_Data.csv, which contains timestamped photovoltaic power generation records. The solar-weather component was derived from SolarPrediction.csv, which contains solar radiation and meteorological variables, including temperature, relative humidity, wind speed, wind direction, and related atmospheric measurements. The final processed dataset is provided as Solar_Weather_Data_Processed.csv. It contains 2,777 continuous hourly samples from 1 September 2016 10:00 to 26 December 2016 02:00. The dataset includes global horizontal irradiance, temperature, relative humidity, wind speed, wind direction, hour, day, month, solar zenith angle, solar azimuth angle, and AC output. The target variable is AC Output, representing photovoltaic AC power output used for supervised regression modeling. The dataset was cleaned and prepared by converting timestamps, resampling hourly, renaming features, converting units, performing feature engineering, calculating solar angles, and final formatting for machine learning experiments. This dataset supports reproducibility of research on explainable and uncertainty-aware solar power forecasting for smart grid applications.本数据集为经筛选与预处理的太阳气象与光伏（Photovoltaic, PV）交流（Alternating Current, AC）功率输出数据集，专为基于机器学习的太阳能功率预测任务打造。本数据集通过整合源自Kaggle平台的公开光伏发电数据与太阳气象数据构建而成。 光伏发电组件数据源自Plant_1_Generation_Data.csv，该文件包含带时间戳的光伏发电量记录。太阳气象组件数据源自SolarPrediction.csv，该文件包含太阳辐射与气象变量，涵盖温度、相对湿度、风速、风向及相关大气观测数据。 最终预处理后的数据集以Solar_Weather_Data_Processed.csv形式提供，包含2016年9月1日10:00至2016年12月26日02:00期间共计2777条连续小时级样本。 该数据集包含全球水平辐照度、温度、相对湿度、风速、风向、时刻、日期、月份、太阳天顶角、太阳方位角以及AC输出等字段。 目标变量为AC Output，即用于监督回归建模的光伏交流功率输出。 本数据集经过了清洗与预处理流程，包括时间戳转换、小时级重采样、特征重命名、单位转换、特征工程、太阳角度计算以及适配机器学习实验的最终格式标准化。 本数据集可支撑面向智能电网应用的可解释性与不确定性感知太阳能功率预测相关研究的可复现性。