A Method Based On Fuzzy Clustering and AHP for Safety Culture Assessment of Chemical Enterprises

Safety culture is a great spiritual prop of enterprises and the soul of safety management, especially for Chemical enterprises, where many incidents have occurred including fire, explosion and poisoning having led to significant numbers of fatalities, injuries and damage. In recent years, some qualitative methods such as interviews, observations, focus groups and action research have been proposed to assess safety culture. The problem of the qualitative methods is that researchers’ subjectivity will affect the assessment results in certain degree. However,it seems that safety culture is difficult to be measured using quantitative methods.

Fuzzy mathematics, a new branch of mathematics, which is used to deal with ambiguity phenomenon in mathematics quantitative methods, is applied to assessing the safety culture of chemical enterprises to solve the above problems in this paper. Six chemical enterprises in Nanjing are chosen to analyze the level of their safety culture. Firstly, a comprehensive and reliable safety culture assessment index system for chemical enterprises is established with the form of target chart including six safety culture indexes that are “safety material culture, safety institution culture, safety behavior culture, safety concept culture, safety culture’s persistence and safety culture’s openness”. Then, the weights of six indexes are determined by applying analytic hierarchy process (AHP), which is the basis of safety culture assessment. Furthermore, by using the fuzzy clustering and MATLAB software platform, a dynamic clustering figure for six chemical enterprises’ safety culture is obtained, and the standard level system of safety culture which is made up of four levels (level1, level2, level3 and level4) is confirmed as well. Finally, the levels of six chemical enterprises’ safety culture are obtained, by comparing the scores of safety culture of six chemical enterprises with the standard scores one by one utilizing the Generalized Hamming distance equation which represents the similarity degree of two quantities.

The classification results of six chemical enterprises’ safety culture got by fuzzy clustering are consistent with the ones got by the Generalized Hamming distance equation. The facts show that the methods used in this paper are reasonable and feasible. Moreover, the use of quantitative method improves the accuracy of assessment results. In fact, the two biggest advantages of the methods used in this paper are as follows: on one hand, the more indexes and samples are chosen, the more accurate results can be obtained; on the other hand, the clustering process is simplified by using MATLAB software. The further studies are needed in the future to build a software system of safety culture assessment for chemical enterprises based on MATLAB platform using the methods in this paper.

Keywords: fuzzy clustering, safety culture assessment, chemical enterprises, analytic hierarchy process