MPC Failsafe Adjustment for Uncertain Scheduled Disturbances
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optimization algorithms for processing plants fail to account for uncertain scheduled disturbances, particularly in plants with discontinuous processes, leading to prediction errors and constraint violations.
Innovation Solution
A computer-implemented method that incorporates a corrective response model with a constraint failsafe parameter to generate optimization adjustment data, accounting for potential worst-case scenarios of scheduled event start times differing from planned times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing optimization algorithms are used for processing plants with discontinuous processes, then the optimization can be performed based on standard assumptions, but prediction errors and constraint violations occur due to uncertain scheduled disturbances
Solution Approach 1:
The patent applies preliminary action by identifying an adjustment value before the scheduled event start time that accounts for potential delays. This adjustment value is incorporated into the corrective response model in advance, allowing the optimization algorithm to proactively compensate for uncertain disturbances rather than reacting after violations occur. The adjustment value represents a worst-case scenario offset that is prepared beforehand to ensure constraints are satisfied even if the scheduled event is delayed.
2Reliability
If optimization algorithms account for uncertain scheduled disturbances, then prediction accuracy and constraint satisfaction improve, but the complexity of the optimization model increases
Solution Approach 1:
The patent applies local quality by making a localized modification to the optimization model only in the specific area affected by scheduled disturbances. Instead of redesigning the entire optimization algorithm, the invention introduces an adjustment value specifically for the corrective response model that accounts for disturbance uncertainty. This localized approach maintains the simplicity of the overall optimization framework while improving reliability only where needed - in the constraint satisfaction for scheduled events.
Solution Approach 2:
The patent applies parameter changes by modifying the corrective response model to include an adjustment value parameter that accounts for scheduled disturbance uncertainty. This parameter is identified before the scheduled event start time and represents a worst-case offset. By changing the model parameters rather than the fundamental algorithm structure, the invention achieves improved prediction accuracy and constraint satisfaction while minimizing the increase in model complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments provide for improved optimization associated with a dynamic control scheme (e.g., an MPC architecture). Some embodiments identify at a first timestamp prior to a planned start time of a scheduled event associated with a processing unit, an adjustment value associated with the scheduled event. The adjustment value may be defined based on a predicted possible worst scenario of starting the scheduled event at a start time that differs from the planned start time. Some embodiments update a corrective response model to include the adjustment value. The corrective response model may comprise a step response matrix, an input vector parameter, and a constraint failsafe parameter. Some embodiments generate, using the corrective response model and based at least in part on operating condition data associated with the scheduled event, an optimization adjustment data, and generate updated optimization data by applying the optimization adjustment data to offset initial optimization data.