A Fitness Distance Correlation-Based Adaptive Differential Evolution for Nonlinear Equations Systems

Nonlinear equations systems (NESs) appear in numerous optimization problems. However, the search performance of the improved evolutionary algorithm is limited due to the lack of analysis of the fitness landscape. In this paper, a fitness distance correlation-based adaptive differential evolution (FDCADE) is proposed. Particularly, firstly FDCADE utilizes fitness distance correlation to extract the local landscape feature of NESs; secondly, according to the fitness landscape correlation, FDCADE adopts the combination of hybrid mutation operators and parameter adaptation strategy to guide the algorithm to search more effectively. To investigate the performance of FDCADE, 18 NESs are used as the test suite. Experimental results demonstrate that FDCADE obtains superior performance on root rate and success rate. In addition, FDCADE also gets competitive results for inverse kinematics of spatial 6R robot manipulators.