In this technique the dynamic relationship between the dependent and the independent variables in form of mathematical models. The MPC algorithm allows for controlling multiple variables simultaneously. Apart from conventional linear MPCs, Tridiagonal also specializes in nonlinear MPCs for accommodating processes that have varying dynamics such as dead-times and lag times.
In this technique, a validated physics-based simulation model is connected with the real system. The simulation models use real-time process data to suggest optimal operating targets. The simulation models are also referred to as digital twins. In order to improve the response time of the simulation models, the physics-based models can be substituted with reduced order statistical models.
In this technique, the readily available process variables such as pressure and temperature are used to predict “hard to measure” process variables (e.g.: stream composition) or product quality. The prediction is done using physics-based or statistical models. Real-time quality prediction enables continuous process optimization.
Although not widely implemented, this option is being widely studied and tested by the entire process industry. Intelligent Process Control allows the engineers to tap the benefits of their historic process data. This technique uses approaches like ANN, fuzzy logic, Bayesian probability and other ML techniques for process control.
