This work is related with human model development for real-time simulation of nuclear power plant (NPP) operator¡Çs cognitive behavior to diagnose plant anomaly through the cntrol interface. The employed method for HUMOS-PAD is largely based on blackboard architecture to configure human information processing mechanism on computer. The information processing is modeled as a number of data-processing subtasks in human working memory (WM) and long-term memory (LTM). Efforts were made to model the inherent variety and diversity in human behaviors. The simulation results demonstrated that the developed model not only can simulate individual operator¡Çs cognitive behaviors well, but also can give good statistical analysis results for human cognitive reliability analysis (HRA) inprobabilistic safety analysis (PSA).