A modified HEART method with FANP for human error assessment in high-speed railway dispatching tasks

Abstract

Human error is recognized as one of the main leading causes of high-speed railway accidents. Human error Assessment and Reduction Technique (HEART) as a human error probability assessment technique has been widely employed in various fields. However, the traditional HEART possess limitations in modelling the dependence relationships among Error Producing Conditions (EPCs) and the uncertainty of experts' estimation in practical application, which may affect its evaluation accuracy. In this paper, a modified HEART method with Railway Action Reliability Assessment (RARA) technique and fuzzy analytic network process (FANP) is proposed to assess human error probability in high-speed railway dispatching tasks. Firstly, the parameters Generic Error Probability (GEP) and Error Producing Conditions (EPCs) from the RARA technique are introduced into conventional HEART technique. Secondly, The FANP is utilized to handle the problems of interdependences and interaction among EPCs and the uncertainty exists in the experts' judgment. The proportion effect of each EPC is determined through ANP by consideration of correlation among EPCs. In addition, the triangle fuzzy number is used to simulate vagueness and uncertainty of expert's opinion. Finally, train occupancy loss in high-speed railway dispatching tasks is selected as an example to illustrate the applicability of the proposed method, and a sensitivity analysis is also performed to test the impact level of the importance degree of experts.