The manned submersible, as the main vehicle of transportation for human navigation in deep sea, has closed and narrow cabin, and with longer operating time, harder operating environment. However, once a manned deep submergence accident occurs, it is highly likely to lead to disastrous consequences. Therefore, it is urgent to develop a dynamic probabilistic risk assessment method which can quantitatively assess the cognitive behavior performance and human error of oceanauts. In this paper, a cognitive behavior simulation method for oceanaut based on knowledge and experience was proposed, which integrated richer performance shaping factors (PIF) into the IDAC model, and a cognitive environment simulation program of the IDAC model was developed through the system dynamics method, so as to analyze the dynamic human error probability of the oceanaut. In addition, the effectiveness and feasibility of the proposed model were evaluated based on real case data of manned deep submergence tasks, and the cognitive characteristics of PIFs were interpreted and classified. The research results are helpful for research institutions of manned deep submersible to accurately identify key risk points, improve the operational efficiency of the human-machine system in the cabin, and reduce the risks of the overall operating system.