Trends in human reliability analysis

Publisher Website
Google Scholar
Abstract
The approach to human reliability has been changing during the past decades, partly due to the needs from probabilistic risk assessment of large scale industrial installations, partly due to a change within psychological research towards cognitive studies. In the paper, some of the characteristic features of this change are discussed Definition of human error and judgement of performance are becoming increasingly difficult concurrently with the change of tasks from routine activities towards decision making during abnormal situations. The nature of human error and the relationship with learning and adaptation are discussed, and the recent development of models of cognitive mechanisms behind errors is mentioned The present approaches to human reliability within different application areas are reviewed. In industrial risk analysis, attempts are made to develop models of operators' decision making during emergency situations, and to obtain the necessary error data by simulator experiments and by systematic use of expert judgement. Simplifying assumptions are necessary for analytical risk assessment including human activities, and to make the results practically acceptable, a close coordination of risk analysis and risk management during operation appears to be necessary. In work safety, the analytical approach of risk analysis seems to be fruitful as a supplement to statistical analysis of accident reports, in particular if supported by application of cognitive models to judge the psychological feasibility of improvements. Finally, an approach to the study of traffic safety from the point of view of intentions and reasons behind behaviour is reviewed and related to the cognitive models described The question is finally raised as to whether the development of cognitive models will be able to serve a more effective transfer of results between these traditionally rather separate lines of research

This publication has 6 references indexed in Scilit: