Stackelberg Game between Charging Stations and Distribution Networks with Regional Load Forecasting and Intelligent Charging Strategies

In order to cooperate with the research of the paper "Stackelberg game between charging station and distribution network based on regional load forecasting and intelligent charging strategy", we have constructed a comprehensive data set, which includes the following contents: README File: This document specifies the computational environment (Python 3.8+, TensorFlow 2.5, Gurobi 9.1), outlines the step-by-step workflow (from data preprocessing and LSTM model training to Stackelberg game simulation), lists key parameter settings, and describes the result verification method. Core Scripts: The bundle includes three Python scripts: (1)data_preprocess.py (performs min-max normalization and splits the load dataset) (2)lstm_load_forecast.py (trains the LSTM model and outputs load predictions) (3)stackelberg_game.py (solves the Stackelberg game using backward induction and outputs DN pricing and CS power purchase decisions) Tiny Synthetic Sample: A 100-hour synthetic dataset, generated based on the statistical distribution of real load data from Tianjin, is provided. It includes features such as date, load demand (kW), dry-bulb temperature (℃), dew point (℃), and hour of day. This sample supports basic LSTM model training and end-to-end verification of the game-theoretic simulation. This data set supports game optimization between distribution network and charging station, load forecasting model training, and simulation and evaluation of intelligent charging strategy for electric vehicles, and is suitable for power system optimization, game theory application and related research of smart grid.为配合论文《基于区域负荷预测与智能充电策略的充电站与配电网斯塔克尔伯格（Stackelberg）博弈》的研究，我们构建了一套综合数据集，具体内容如下： README文件：本说明文档明确了计算环境（Python 3.8+、TensorFlow 2.5、Gurobi 9.1），梳理了从数据预处理、长短期记忆网络（Long Short-Term Memory, LSTM）模型训练到斯塔克尔伯格博弈模拟的完整分步工作流程，列出了关键参数设置，并详细说明了结果验证方法。 核心脚本：本脚本包包含三个Python脚本： (1) data_preprocess.py：实现最小-最大归一化处理并拆分负荷数据集 (2) lstm_load_forecast.py：训练长短期记忆网络（LSTM）模型并输出负荷预测结果 (3) stackelberg_game.py：采用反向归纳法求解斯塔克尔伯格博弈，输出配电网定价与充电站购电决策 小型合成样本：提供了一份基于天津地区真实负荷数据统计分布生成的100小时合成数据集，其包含日期、负荷需求（单位：kW）、干球温度（单位：℃）、露点温度（单位：℃）以及当日时段等特征。该样本可支持基础LSTM模型训练以及博弈论模拟的端到端验证。 本数据集可用于配电网与充电站间的博弈优化、负荷预测模型训练，以及电动汽车智能充电策略的仿真与评估，适用于电力系统优化、博弈论应用及智能电网相关研究领域。