As an important equipment for the safe and stable operation of nuclear power plants, the human-machine interface in the main control room becomes possible to be intelligently designed with the development of multi-modal interaction technology. This paper studies the intelligent human-machine interaction design of the main control room of nuclear power plants based on the method of task analysis. By selecting typical scenarios, the monitoring and control tasks of operators under different operating conditions of the power plant were analyzed in detail, and the human engineering discrepancies(HED) related to the human-machine interface were identified, such as the need to optimize the displays and procedures, as well as the problem of high workload of operators. To address the issue of high workload for operators, this paper proposes a solution that introduces intelligent voice interaction technology. By designing the architecture and key technologies of the intelligent voice interaction system based on the functional requirements derived from task analysis, it is verified in a laboratory environment that the intelligent voice interaction system can effectively reduce the workload of operators and improve the efficiency of task execution, solving the human engineering discrepancies found in task analysis. This paper provides new ideas and methods for the human-machine interaction design of the main control room of a nuclear power plant.