A visualization method for operating optimization

This paper proposes a visualization method for finding the optimal operating region and a near-optimal operating point based on process data. The basic principle behind the method is to map data in multi-dimensional space to a two-dimensional plane, and to generate simultaneously the contours of the objective function or functions with a mapping model. The optimal operating region can be located intuitively according to the contour distributions of the objective functions and a point found which is in this region and although not strictly optimal is near-optimal. The optimal point found can be mapped back to the original multidimensional space with a reverse mapping method proposed previously. The visualization method is tested on four operating optimization problems, including problems with multiple objective functions. Results show that the proposed method is able to reveal fully the features of the operating space and to find effectively the optimal operation region and a good operating point.