Based on human model simulation, an estimation method for basic Human Error Probability(HEP) parameters is presented, for Probability Safety Assessment (PSA) / Human Reliability Assessment(HRA) in Nuclear Power Plant(NPP). A framework of HEP parameters needed for PSA/HRA is firstly described, and then in order to construct a human model on plant anomaly diagnosis at NPP, methods of laboratory experiment are briefly explained. An experiment was conducted for obtaining information about operators¡Ç cognitive behavior. Two kinds of Human Cognitive Reliability(HCR) curves for detecting and diagnosing an anomaly were derived from the experimental data and the characteristics of both curves were analyzed with respect to the nature of time dependency of the HEP. The human model was constructed for simulating the operators¡Ç cognitive behaviors during the detection and diagnosis phase. Using the human model, HCR curves for anomaly detection phase were deduced by computer experiments. The inter-comparisons of the HCR curves with the experimental data resulted in that the application of human model simulation would be feasible for the estimation of HEP parameters needed for PSA/HRA.