Codes underlying: Comparison of scenario reduction approaches for reservoir inflow scenarios generated by a Bayesian Neural Network

The data set and codes for a paper, Comparison of scenario reduction approaches for reservoir inflow scenarios generated by a Bayesian Neural Network.Including reservoir inflow data for the Daecheong reservoir in South Korea, there are codes to build a BNN model with hyperparameter optimization using the TPE algorithm. In addition, codes for scenario reduction by four different measures, Wasserstein, energy, Euclidean, and Manhattan distances, are integrated.

本数据集与代码配套于一篇题为《贝叶斯神经网络生成的水库入库情景的情景削减方法对比》的学术论文。数据集涵盖韩国大清水库的入库径流数据，同时提供可通过TPE算法实现超参数优化以构建贝叶斯神经网络（Bayesian Neural Network）模型的代码。此外，本资源还集成了基于瓦瑟斯坦距离（Wasserstein）、能量距离（energy）、欧几里得距离（Euclidean）与曼哈顿距离（Manhattan）四种不同距离度量的情景削减代码。