Continuous Time Stochastic Modelling means semi-physical modelling of dynamic systems based on stochastic differential equations. Stochastic differential equations contain a diffusion term to account for random effects, but are otherwise structurally similar to ordinary differential equations. This means that conventional modelling principles can be applied to set up the model.
CTSM provides methods for estimating unknown parameters of the model from data, including parameters in the diffusion term. These methods are able to handle both linear and nonlinear models, and CTSM also provides flexibility with respect to the data sets that can be used for estimation, e.g. by allowing varying sample times, missing observations and occasional outliers.
The parameter estimation methods implemented in CTSM are a maximum likelihood (ML) method and a maximum a posteriori (MAP) method. Both methods allow several independent data sets to be used. The ML and MAP methods are both sound statistically based estimation methods, which means that once the parameters have been estimated, statistical methods can be applied to investigate the quality of the model. Features that facilitate application of such methods are also included in CTSM.