Plant Cycle Optimization Using Two-Stage Data Record Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cyclic processes in the basic-materials industry, such as steel and aluminum production, exhibit significant deviations from cycle to cycle, leading to suboptimal operation due to the complexity of influencing variables and opposing parameters, resulting in inefficient attainment of key performance targets like productivity and energy minimization.
Innovation Solution
An optimization method executed by a computer that utilizes a model of the plant and process to ascertain expected values for target variables, selects data records with minimal distance from these values, and outputs set values to operators or control devices to achieve optimal process control, considering both historical and dynamic data records.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a model-based optimization method is used to ascertain expected values for target variables, then the attainment of key performance targets is improved, but the complexity of the optimization system increases
Solution Approach 1:
The patent segments the optimization process into two distinct stages: a first selection stage that filters data records based on first target variables and first actual variables, and a second selection stage that further filters based on second target variables and second actual variables. This segmentation reduces the complexity of simultaneously optimizing multiple conflicting parameters by breaking down the complex multidimensional optimization problem into sequential, manageable filtering steps.
2Productivity
If multiple target variables are optimized simultaneously, then the overall process efficiency is improved, but the difficulty of controlling opposing parameters increases
Solution Approach 1:
The patent implements a dynamic, two-stage selection process where the criteria for data record selection are adjusted in sequence. The first stage optimizes for certain target variables, and the second stage refines the selection based on additional target variables. This dynamic approach allows the system to handle opposing parameters by adapting the selection criteria at different stages, making the control of multiple target variables more manageable.
3Reliability
If historical data records are used for optimization, then the reliability of process control is improved, but the time required for data processing increases
Solution Approach 1:
The patent performs preliminary filtering of historical data records in the first selection stage based on first target variables and first actual variables. This preliminary action reduces the volume of data that needs to be processed in the subsequent second selection stage, thereby reducing the overall data processing time while still utilizing historical data to improve the reliability of process control.
4Manufacturing precision
If strict selection criteria are applied to data records, then the manufacturing precision of process outcomes is improved, but the quantity of available data records decreases
Solution Approach 1:
The patent applies partial selection criteria in a two-stage process. The first stage applies filtering based on first target variables and first actual variables, and the second stage applies additional filtering based on second target variables and second actual variables. This partial, sequential application of selection criteria maintains manufacturing precision by ensuring data records meet multiple criteria, while avoiding the need to apply all strict criteria simultaneously, which would eliminate too many data records.
Data Source
AI summary
An optimization method in which a computer ascertains expected values (E1) for actual variables (I1) of a technical process based on values (R) for target variables (Z1) of the technical process that attain the values (R) as far as possible. From data records (D), the computer provisionally selects a number (n1) of records (D) in which the variables (I1) display a minimum distance from the values (E1). The computer then ascertains expected values (E2) for the actual variables (I2) based on the values (R) and the values (E1). From the provisionally selected data records (D), the computer selects a predetermined second number (n2) of data records (D) in which the variables (I1, I2) display a minimum distance from the values (E1, E2). The computer ascertains set values (S) for the variables (Z2) for a yet-to-be-executed cycle to attain variables (Z1) as close to possible to the values (R).