| Budget Amount *help |
¥3,200,000 (Direct Cost: ¥3,200,000)
Fiscal Year 2001: ¥1,300,000 (Direct Cost: ¥1,300,000)
Fiscal Year 2000: ¥1,900,000 (Direct Cost: ¥1,900,000)
|
|---|
| Research Abstract |
The object of this research is to establish a method to identify root causes of a system accident using the knowledge on human errors obtained in human factors, and cognitive engineering etc.
At the first stage in Heisei 12, a framework was considered to identify root causes of human errors and component failure. To estimate the human cognitive process, a concept of 'mental images (MI)' was applied. From the viewpoint of how to recognize situations (generation of MI), how to make a decision on what to do (action planning based on MI), and how to act (execution based on MI), errors are identified as deviations from the desired criterion. Application to marine accidents showed that root causes were obtained and the merits of the method were confirmed. Further, as the situation awareness was essential, how to allocate monitoring points was derived to identify the system state. Related to system state identification, how to evaluate the effect of a potential component failure on the system function was also considered.
In the second stage in Heisei 13, the analysis method to identify root causes of human errors was improved, and further problems to be solved were clarified. Based on MI, the surface human errors were classified into errors in acquisition of input information, errors in judgment, and errors in action execution, Based on MI formation from input imformation sources, the framework was improved to identify root causes of each type of errors. By applying the proposed method to a ship accident, its merits were confirmed. Further, a derivation method of accident sequences was proposed to consider safety measures to prevent accidents in a complex system. The method is based on a global system model composed of hardware behavior such as components, software behavior such as computerized actions, and human actions.
|
