A Self-Adaptive Response Strategy for Dynamic Multiobjective Evolutionary Optimization Based on Objective Space...

Area of Science:

• Optimization Algorithms
• Evolutionary Computation
• Dynamic Systems

Background:

• Dynamic multiobjective optimization (DMOP) involves optimizing multiple conflicting objectives that evolve over time.
• Existing response strategies struggle with diverse environmental changes in DMOPs.
• The need for adaptive algorithms to address unknown environmental dynamics is critical.

Purpose of the Study:

• To propose a new dynamic multiobjective evolutionary algorithm (DMEOA) capable of handling unknown environmental changes.
• To enhance the tracking performance of DMEOAs in dynamic environments.
• To apply the proposed algorithm to real-world control problems, such as PID controller tuning.

Main Methods:

• Developed a DMEOA utilizing objective space decomposition.
• Incorporated a maxi-min fitness function for selection.
• Designed a self-adaptive response strategy integrating multiple response mechanisms.
• The strategy adaptively selects response methods based on past performance.

Main Results:

• The proposed DMEOA demonstrates competitive performance across various dynamic multiobjective optimization problems (DMOPs).
• The self-adaptive strategy effectively manages unknown environmental changes.
• The algorithm achieved superior control effects when applied to PID controller parameter tuning for dynamic systems.

Conclusions:

• The novel DMEOA with a self-adaptive response strategy is a promising approach for tackling DMOPs with unknown environmental dynamics.
• The algorithm offers improved tracking and adaptation capabilities compared to existing methods.
• Its successful application to PID controller tuning highlights its practical utility in dynamic system control.