A new measure on adaptation complexity— fitness function classes, their integration and case study

How to effectively measure the adaptation complexity is an open issue in nature-inspired computation. In this paper, some essential characteristics of adaptation in evolution and the importance/complexity of constructing multi-objective fitness functions in evolutionary computation are analyzed. Based on the author’s former work on the single-objective normalization, a general method is brought forward for multi-objective decision making and optimization whose key point is to divide the process of constructing fitness functions into there basic cases. Then the issues on the determination of the corresponding mathematical models and their parameters, the integration of all the fitness functions into a multi-objective fitness function are discussed. A paradigm in multi-input-multi-output control systems is illustrated to show the technical route of our method.