Non-measurable process variables are estimated by combination of plant measurements and an adaptive process model.
Cybernetica ModelOnline – Soft sensing based on mechanistic models
Cybernetica ModelOnline combines a process model and an estimator algorithm to provide on-line estimation (soft sensing) of process variables that cannot be directly measured, such as:
- Dynamic data reconciliation (e.g. estimation of flow rates in oil and gas production).
- Reaction rates for chemical reactions.
- Product quality parameters.
- Concentrations of raw materials, reaction products and other chemical species.
- Conversion of chemical reactions in batch, semi-batch or continuous processes.
- Important process parameters, such as catalyst activity or heat transfer coefficients for cooling utilities.
Model and application development
Process models as well as application specific codes for control of the estimator algorithms are implemented in a Cybernetica Model and Application Component, which is linked into the ModelOnline system. This separation allows for very specific tailoring in order to best meet our customer needs, while we still build the total application upon a general software kernel and a collection of advanced algorithms for estimation. Cybernetica Model and Application Component features:
- Models can be coded directly in the C programming language.
- Models can be developed in Modelica or other modelling tool that supports the Functional Mockup Interface (FMI) standard.
- Analytical Jacobians are supported.
- ODE/DAE solvers with sensitivity integration are supported.
Cybernetica ModelFit is used for off-line estimation of model states and parameters, for model validation, and for design of the on-line estimation in Cybernetica ModelOnline applications. ModelFit is used to decide which model parameters should be estimated on-line, to design the on-line estimators, and to estimate the parameters that are considered constant.