In this paper, we describe how to optimize multiple objective functions in uncontrollable environment changes. We use Bayesian optimization to optimize for a changing environment in complex systems where the form of the objective function is unknown. The difference from the previous study is that the controlled input values and the given environmental values are selected in a continuous range rather than in a set. To do this, we define a virtual Pareto set using the predictive distribution of the Gaussian process and present a CMOBO algorithm. The proposed algorithm describes how optimization is performed when wind direction is changed for the wind farm data generated by the FLORIS simulator.