Engine Model Parameter Optimization via Reference Point Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-dimensional engine parameter optimization is challenging due to numerous non-dominated solutions and low selection pressure caused by high dimensionality, as well as dependence on reference point sets, which complicates the optimization process and increases computational waste.
Innovation Solution
An optimization method that sets an initial population of engine model parameters, evolves it through selection, crossover, and mutation based on fitness values, and updates the population until a termination condition is met, avoiding the need for reference point sets and simplifying the optimization process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If dominance relationship-based multi-target optimization methods (such as NSGAII and SPEA2) are adopted, then multi-target optimization can be achieved, but the number of non-dominated solutions becomes large and selection pressure becomes low due to high dimensionality
Solution Approach 1:
The patent divides the multi-target optimization problem into multiple single-target optimization problems by introducing reference point sets. Each reference point represents a target direction, and the optimization process evaluates solutions based on their performance relative to these reference points. This segmentation allows the algorithm to maintain focused selection pressure on each target dimension independently, avoiding the dilution of selection pressure that occurs in high-dimensional multi-target problems.
Solution Approach 2:
The patent changes the optimization approach by introducing reference point sets with specific mathematical properties (norm constraints, normalization). By transforming the multi-target problem into a reference-point-based evaluation framework, the algorithm can effectively handle high-dimensional objectives while maintaining adequate selection pressure through the structured reference point geometry.
2Adaptability or versatility
If decomposition-based multi-target optimization methods (e.g., MOEA/D) are adopted, then multi-target optimization can be achieved, but the number of calls to the engine model increases when using larger populations
Solution Approach 1:
The patent segments the optimization evaluation by introducing reference point sets that define specific target directions. Instead of evaluating all objectives simultaneously for each individual in the population, the algorithm evaluates solutions against these segmented reference points. This allows for more efficient population management and reduces the total number of expensive engine model calls required.
Solution Approach 2:
The reference point set serves multiple functions: it defines target directions, provides evaluation criteria, and enables population management. By making the reference point system multi-functional, the patent reduces the need for separate mechanisms for each function, thereby reducing overall computational overhead.
3Device complexity
If conversion to single-target problem is adopted, then optimization can be simplified, but conflicts among targets arise and overall target value needs to be reduced at the expense of performance in some dimensions
Solution Approach 1:
The patent segments the multi-target optimization into a reference point-based framework where each reference point represents a target direction. This segmentation allows the algorithm to handle multiple targets simultaneously without converting to a single-target problem, thereby avoiding target conflicts while maintaining a manageable optimization process through the structured reference point evaluation.
Data Source
AI summary
An optimization method, apparatus, computer device and storage medium for engine model. The method includes: setting an initial population according to a predetermined number of solutions; obtaining a first output set of the initial population on the engine model; determining a first fitness value; operating selection, crossover, and mutation on the initial population according to the first fitness value to obtain an evolved population, and the initial population and the evolved population constitute a to-be-selected population; obtaining a second output set of the evolved population on the engine model; determining a second fitness value; selecting engine model parameters in the to-be-selected population to form a new population according to the second fitness value; and optimizing the engine model by using the engine model parameters in the new population if a termination condition is met.