A novel evaluation model based on connection cloud model and game theory under multiple uncertainties

Yan Wang

Abstract: Under uncertainty, the evaluation inevitably involves the randomness and fuzziness of indexes and weighting. Although the normal cloud model can depict the fuzziness and randomness of the index, it ignores the changing tendency in ranking. It requires the index subjecting a normal distribution that may not match the actual state of the indicator distribution. A novel evaluation model based on the connection cloud and game theory was presented to address multiple uncertainties. In this model, the stochastic and fuzzy uncertainties of evaluation indexes distributed in the finite intervals are first simulated for the interval-valued evaluation standard by the connection cloud model. Then, the combined weight is assigned based on the game theory considering the harmony between the conflicting subjective and objective weights. The aggregated value of the connection degrees is further specified to describe the certainty-uncertainty relationship between the sample and evaluation standard of each rating level and the conversion tendency of the evaluation from a unified view dialectically. Furthermore, a case study on the pavement performance evaluation and comparisons with other models confirmed the feasibility and validity of the proposed model. The results show that the evaluation model presented here can effectively depict actual distribution characteristics and multiple uncertainties of indexes and thoroughly explore the inherent information importance.