Continuous Time Stochastic Modelling for R or CTSM-R is a free, open source and cross platform tool for identifying physical models using real time series data. Visit ctsm.info for all information and how to install CTSM-R on your computer. Download the user guide from here. Both non-linear and non-stationary systems can be modelled using CTSM-R. By using a continuous time formulation of the dynamics and discrete time measurements the framework bridges the gap between physical and statistical modelling.
CTSM-R is the latest incarnation of CTSM which itself can be traced back more than 30 years. CTSM-R extends the free and open source statistical platform R (www.r-project.org) thus combining the power of CTSM and the data handling, plotting and statistical features of R. CTSM was partly developed during the EU PASSYS project (1 April 1986 to 31 March 1989) and a number of follow up research project funded by EU.
CTSM-R has been successfully applied to a wide range of data-driven modelling applications: heat dynamics of walls and buildings, dynamics of heat exchangers, radiators and thermostats, solar thermal collectors, building integrated photovoltaic systems and more.
It is possible to generate both pure simulation and k-step prediction estimates of the states and the outputs, filtered estimates of the states and, for nonlinear models, smoothed estimates of the states. Standard deviations of all state and output estimates are now also automatically provided. The CTSM modelling framework includes methods for model validation and for selecting the best model among a class of candidate models.