Genetic Algorithm Forecasting with Process Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing forecasting methods using genetic algorithms often generate forecasts that do not account for real-world process constraints, leading to unrealistic predictions.
Innovation Solution
A computer-implemented method and system that uses a genetic algorithm to generate chromosomes with data values for equations, calculates chromosome values based on a goal function, and compares process parameter values to constraints, modifying the chromosome values if they do not satisfy the constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If mathematical analyses are used to generate forecasts based on historical data, then statistical accuracy is improved, but the forecasts fail to account for process constraints leading to unrealistic predictions
Solution Approach 1:
The patent modifies the optimization approach by changing the parameters being optimized from purely statistical fit to constraints-based feasibility. The genetic algorithm optimizes chromosome values (equation parameters) subject to process constraint boundaries, transforming the forecast generation from unconstrained statistical interpolation to constrained physical feasibility optimization.
Solution Approach 2:
The patent implements feedback through iterative evaluation of process constraints during the genetic algorithm optimization. Each chromosome is evaluated against process constraints, and chromosome values are modified based on constraint violations, creating a feedback loop that ensures forecasts remain physically realistic while maintaining statistical accuracy.
2Measurement precision
If genetic algorithms are used to generate forecast equations, then statistical fit is improved, but the forecasts violate real-world process constraints
Solution Approach 1:
The patent applies preliminary action by pre-establishing process constraint boundaries before running the genetic algorithm. These constraints are encoded into the optimization process, ensuring that chromosome values (equation parameters) are guided toward feasible regions from the outset, preventing generation of physically impossible forecasts.
Solution Approach 2:
The patent uses process constraints as intermediary elements that mediate between statistical accuracy and physical feasibility. The constraints act as intermediate filters that evaluate and modify chromosome values, ensuring that the optimization process balances statistical fit with real-world manufacturability and process feasibility.
Data Source
AI summary
A characteristic forecasting system is disclosed. The characteristic forecasting system may have a memory and a processor. The memory may store instructions, that, when executed, enable the processor to generate at least one chromosome using a genetic algorithm, the chromosome including data values for variables of one or more equations used to generate forecast data for a target item. The processor may also be enabled to calculate a chromosome value for the chromosome based on a goal function associated with the genetic algorithm and determine at least one process parameter value for the chromosome at a time interval of the forecast data. The processor may also compare the process parameter value to a process constraint value representing a process limitation associated with the target item and modify the chromosome value for the chromosome responsive to a determination that the process parameter value does not satisfy the process constraint value.