Real-Time Process Simulator Using Segmented Linear Models

The patent segments the complex non-linear multivariable model into multiple linear models, each representing a specific operating point or region of the process. These segmented linear models can be solved quickly and individually, avoiding the computational burden of solving a single complex non-linear model while maintaining acceptable accuracy through piecewise approximation.

The patent implements a dynamic model selection mechanism that automatically transitions between different linear models based on current process conditions. This dynamic approach allows the system to adapt to changing operating conditions while maintaining real-time computational efficiency, as each linear model can be evaluated and switched between rapidly compared to re-solving a complex non-linear model.

The patent divides the complex non-linear modeling task into multiple simpler linear modeling tasks. Each linear model can be derived using standard, well-established techniques rather than requiring complex non-linear model derivation. This segmentation reduces the expertise and resources needed while collectively capturing the essential non-linear behavior through multiple linear approximations.

The patent changes the mathematical form of the models from non-linear to linear, which fundamentally simplifies the derivation process. Linear models can be derived using conventional system identification techniques and require less computational effort for parameter estimation. The collective behavior of multiple linear models with different parameters approximates the non-linear behavior without requiring complex non-linear mathematics.

The patent merges multiple linear models into a unified framework that collectively represents non-linear behavior. By combining several linear models with different operating characteristics, the system achieves both computational efficiency (inheriting from linear models) and non-linear simulation capability (emerging from the combination). The merged model structure allows real-time computation while capturing non-linear effects through the ensemble of linear models.

The patent creates a universal modeling framework where linear models serve multiple functions: they provide computational efficiency for real-time simulation, enable easy derivation through standard techniques, and collectively represent non-linear behavior. This multi-functionality resolves the contradiction by making linear models versatile enough to replace both simple separate simulators and complex non-linear models depending on the application needs.