MINLP Constraint Grouping for Real-Time Process Optimization

The patent segments the MINLP problem into two distinct phases: a discrete event simulation phase that handles switchable units and operational constraints, and a continuous optimization phase that handles smooth nonlinear programming. This segmentation allows each phase to be solved with appropriate methods, reducing overall computational complexity and time while maintaining solution optimality.

The patent performs preliminary discretization of the time horizon and preliminary identification of switchable units and their operational constraints before the main optimization. This preliminary action prepares the problem structure in advance, allowing the solver to focus computational effort on the critical optimization decisions rather than spending time on basic setup and constraint formulation during the solving process.

The patent implements a dynamic solution approach where the optimization problem is reformulated to separate discrete switching decisions from continuous operational parameters. The discrete events are handled through event-driven simulation logic that dynamically adjusts the continuous optimization intervals, allowing the computational system to adapt its resource usage to the actual problem complexity rather than allocating resources for all possible scenarios.

The patent substitutes the traditional mechanical MINLP solver approach with a hybrid methodology that replaces parts of the mathematical programming mechanism with discrete event simulation logic. This substitution eliminates the need for the solver to handle complex integer constraints and switchable unit logic, reducing computational burden on processors and memory while maintaining accurate determination of optimal switching decisions.