A Novel Risk Matrix Approach Based on Cloud Model for Risk Assessment Under Uncertainty

The fuzziness and randomness are significant epistemic uncertainty within the qualitative categorization of the two critical factors, the frequency and severity, which are not fully considered in the traditional risk matrix. This paper mainly proposes a cloud risk matrix method for the risk assessment of process safety considering the epistemic uncertainty in expert elicitation. The cloud model is employed to provide a mathematical expression for the fuzziness and randomness in the linguistic variables by its two numerical characteristics entropy En and hyperentropy He. An adjusted Mamdani inference algorithm is constructed for the determination of an integrated risk cloud given the value of input variables. And the centroid method is improved to be adapted to the calculation of risk index from the enormous cloud droplets in the integrated risk cloud. A case study of risk assessment for distillation column unit is performed to illustrate the process of cloud risk matrix in detail, and the validity and rationality are verified by contrast to the results from the fuzzy risk matrix. Results indicate the proposed method can handle the randomness of qualitative concepts, which is more suitable to the practical condition. Moreover, the effect of the hyperentropy He on the randomness of risk index is also investigated and discussed for the reference to parameter selection. The proposed cloud risk matrix can provide an effective risk inference tool in a wide range of engineering fields.