We provide two datasets gengine1 and gengine2. Each dataset consists of N measurements of the gasoline engine D = {d1,...,dN}. Each measurement d = {(u1, y1),...,(uT, yT)} contains a trajectory of length T where the inputs are denoted as ut and the outputs as yt. The length of the trajectory may vary between measurements. The data points are recorded every 10 Hz. Each measurement is saved as a separate csv-file. The data is standardized to mean-zero and unit-variance on the training data set.
The dataset gengine1 contains in total 33 measurements. We used the measurements [10,11,12,13,14,15,16,17,18,19,30,31,32,33,34,35,36,37,38,39] for training and the measurements [53,54,55,56,57,58,59,60,61,62,63,64,65] for testing.
The first 5 columns correspond to the inputs ut:
- Speed
- Load
- Lambda
- Ignition Angle
- Fuel Cutoff
The last 8 columns correspond to the outputs yt:
- Particle Numbers
- CO
- CO2
- HC
- Nox
- O2
- Temperatue (Exhaus Manifold)
- Temperate (Catalyst 1)
The dataset gengine2 contains in total 22 measurements. We used the measurements [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15] for training and the measurements [16,17,18,19,20,21] for testing.
The first 4 columns correspond to the inputs ut:
- Speed
- Load
- Lambda
- Ignition Angle
The last 5 columns correspond to the outputs yt:
- Particle Numbers
- HC
- Nox
- Temperatue (Exhaus Manifold)
- Temperature (Catalyst 1)