A dynamic industrial design optimization process requires high-quality optimization algorithms as well as adaptive representations capable to find the global solution for a given problem. To adapt the representation to changing environments or new input we utilize the concept of evolvability. Our interpretation of this concept consists of three criteria, namely variability, regularity, and improvement potential, where regularity and improvement potential characterize conflicting goals between exploration and exploitation. We target the efficient realization of the adaptation process of the representation according to a given preference weight between regularity and improvement potential. We propose a combination of two heuristics, Lloyd and orthogonal least squares sampling, to initialize the adaptation process of the representation for a given preference. This initialization improves the convergence speed of the adaptation process and the fitness of the results. Then, we realize a stepwise design optimization procedure, where we alternate the adaptation process of the representation with the design optimization. During the design optimization process we extract information which we use in the following adaptation phase. We show that an intermediate preference between regularity and improvement potential utilizes this information as well as overcomes erroneous initial information. Thereby, we increase the performance of the whole design optimization process.
