Observer module¶
The HILO-MPC Observer module contains several state (and parameter) observers. Observer are used to infer states and parameters from measurements. For a more detailed description of the methods refer to the API.
The Observer module contains the following classes:
Moving Horizon Estimator (MHE)
Kalman Filter (KF)
Extended Kalman Filter (EKF)
Unscented Kalman Filter (UKF)
Particle Filter (PF)
Moving Horizon Estimator¶
Non-uniform sampling intervals¶
Multi-rate measurements¶
Kalman Filter¶
The class KalmanFilter (alias KF) implements the Kalman filter developed by Rudolf E. Kálmán. To set up the Kalman filter you need an already set up Model instance. Additionally you might want to supply a plot backend (via the plot_backend keyword argument) in order to visualize the estimation results later on. At the moment only Matplotlib and Bokeh are supported for plotting. The Kalman filter can be initialized as follows:
from hilo_mpc import KF
# Initialize Kalman filter
kf = KF(model, plot_backend='bokeh')
Required information, like e.g. the model dynamics or the sampling time, will be automatically extracted from the Model instance.