Supplementary dataset for paper: "Approximate non-linear model predictive control with safety-augmented neural networks"

Supplementary dataset for paper Henrik Hose and Johannes Koehler and Melanie N. Zeilinger and Sebastian Trimpe "Approximate non-linear model predictive control with safety-augmented neural networks". The code to use this dataset is publicly available at https://github.com/hshose/soeampc The dataset contains training and testing data to train an NN controller for three standard benchmark systems, a stir tank reactor, a quadcopter, and a chain mass system. For each system, there are initial conditions as comma separated value in the `x0.txt` file, the MPC input trajectory in the `U.txt` file and the corresponding predicted state sequence in the `X.txt` file. MPC parameters are provided for each system. The dataset was computed using acados for SQP solving. The dataset also contains pretrained neural network approximations of the dataset.These are provided in the `pretrained_models.zip` file. The neural networks were trained with tensorflow.本数据集为Henrik Hose、Johannes Koehler、Melanie N. Zeilinger与Sebastian Trimpe所著论文《Approximate non-linear model predictive control with safety-augmented neural networks》（近似非线性模型预测控制结合安全增强神经网络）的补充数据集。该数据集的使用代码已公开于https://github.com/hshose/soeampc。本数据集包含用于训练神经网络控制器的训练与测试数据，覆盖三类经典基准系统：搅拌槽反应器（stir tank reactor）、四旋翼飞行器（quadcopter）以及链式质量系统（chain mass system）。针对每个系统，`x0.txt`文件以逗号分隔值（Comma-Separated Values, CSV）格式存储了初始条件，`U.txt`文件存储了模型预测控制（Model Predictive Control, MPC）输入轨迹，`X.txt`文件存储了对应的预测状态序列。同时为每个系统提供了模型预测控制相关参数。本数据集通过acados工具完成序列二次规划（Sequential Quadratic Programming, SQP）求解计算。数据集还包含针对该数据集的预训练神经网络近似模型，存储于`pretrained_models.zip`压缩包中。上述神经网络采用TensorFlow框架训练得到。