A Cross-Evaluation Matrix (CEM) Rating Based Approach to Pareto Efficient Frontier and Design Alternatives Enumeration

This paper introduces a novel cross-evaluation matrix (CEM) rating based approach for the development of Pareto efficient frontiers and the finding of discrete sets of globally non-inferior design sets. This work, based on concepts from data envelopment analysis (DEA), can facilitate the enumeration of design candidates in a multi-criteria formulation. In addition, it is expected that the resulting design sets will provide the basis for the establishment of a value system and subsequent preference based rank ordering of expected outcomes in the single-criterion formulation. A unique feature of this cross-evaluation matrix approach is its ability to handle problems without requiring a priori tradeoff formulation or multiattribute model development. As such, its application does not require assignment of a set of a priori weight constants as in many well-established Pareto-optimal generating methods, nor does it need any a priori information of the global minimum or maximum of the attribute functions. Recognizing that the enumeration of multiple discrete solution alternatives can be best achieved in a parallel computation environment, the implementation in this work is executed with the aid of a genetic algorithm strategy. The effectiveness of the integrated approach in yielding Pareto-optimal candidate design sets under different scenarios are studied in the context of illustrative examples, including two engineering case studies, and the results are discussed.