With the increase of public environmental and safety awareness, the safety issue incomplex safety-critical systems such as nuclear power plant(NPP) is paid more andmore attention to. Human reliabilty plays an essential role in such safety-criticalsystems. With the introduction of computerized operation systems, the task typesbecome more challenging for operators. Research on the effects of task complexity andtime pressure on operator’s performance is a meaningful topic for improving HumanReliability Assessment(HRA), crew training and design of operation procedures.Thus, an experiemental study is carried out in this thesis, including the followingaspects:(1) Using the TC (Task Complexity) method to measure task complexity anddesign procedurized tasks. In this method, each step of the procedure is splitted into themost basic recognizable units, namely meta-operations. Each meta-operation can beanalysed from one of the four cognitive resources, i.e., Visual, Auditory, Cognitive andPsychomotor Channels (VACP). The quantity of resources used by each meta-operationis given by the VACP scale. Thereby in this way, task complexity is measured (TCScore).(2) Based on the complexity measurement, a simulated Emergency OperationProcedures (EOP) with the NPP background was constructed with15levels of TCScore in a relative uniform distribution with a wide range. In the VB.Net2010environment, an experiement platform was developed with monitoring and EOPinterfaces. A multi-display operation experiement was conducted with74participants.(3) After collecting the performance data of those74participants, regression analysiswas done to explore models between TC Score and performance under five differenttime pressure levels and two proficient stages. The performance variaables include:average operation time, error rate, error type, perceived difficulty and perceivedcomplexity of each step.The results from the experiement indicate significant effects of TC Score onperformance. However, the regression models of TC Score vs. performance relationshipturn out different under different time pressure levels and different proficient stages.Specifically, the TC Score vs. Operation Time models are all linearly significant, whilethe TC Score vs. Error Rate relationship is only linearly significant under high time pressure conditions and it should be fitted into other models under low time pressureconditions. The distribution of error type shows no significant difference between timepressure levels, but varies a lot for different steps (complexity levels). Perceiveddifficulty and perceived complexity were measured respectively before and after thetraining stage of the experiement. They are all linearly related with TC Score, whichverifies the rationality of the TC method for measuring task complexity. It was foundthat post-perceived complexity can better predict error rate. |