The Research On Evaluation Of Digital Humanmachine Interface In Nuclear Power Plant Based On Elman Neural Network

The main control room of a nuclear power plant is the central hub for monitoring and controlling the operating status of the plant.The quality of the digital human-machine interface design is crucial for the safe operation of the nuclear power plant.As the main control rooms transition from analog to digital instrument control,the risk of human error increases even as operators are provided with better support.By assessing the human-machine interface,the design can be improved,operator workload can be reduced,and the probability of human errors can be minimized.Hence,conducting research on the evaluation of digital human-machine interface in nuclear power plants has significant theoretical and practical value.Presently,research on human-machine interface evaluation in nuclear power plants mainly focuses on subjective evaluations of backup panel interfaces,static interfaces,and layout optimization,with limited consideration for human factors in human-machine interaction.In this paper,the digital human-machine interface of a pressurized water reactor nuclear power plant’s main control room was chosen as the research subject.An evaluation index system,based on the features of the digital human-machine interface and humanmachine interaction,was constructed.The grey relational fuzzy clustering method was applied for clustering optimization of the evaluation index system.Weight allocation for multiple indices was achieved using the grey distance TOPSIS method,and scores for each interface and evaluation factor were calculated using expert subjective evaluation data.An integrated evaluation model of the digital control interface of nuclear power plants based on the Elman neural network was subsequently developed.To address issues related to unactivated hidden layer nodes and low convergence accuracy in the Elman neural network evaluation model,the particle swarm optimization algorithm was employed for model optimization.Differential evolution and simplex algorithms were utilized to improve the global and local search capabilities of the particle swarm optimization algorithm,resulting in the formation of the DE-PSO,SM-PSO,and hybrid particle swarm algorithms.These improved particle swarm optimization algorithms were combined with the Elman neural network evaluation model.It was found that the hybrid particle swarm Elman evaluation model demonstrated higher convergence accuracy,faster convergence speed,and more accurate evaluation results.An example of the reactor startup procedure,starting from criticality to 100% rated power output,was employed for the comprehensive evaluation of the digital human-machine interface.The developed software for evaluating the digital human-machine interface was used to assess the current interface,provide optimization suggestions according to the evaluation results,and design a function-centered digital control interface based on the results and digital procedures.The evaluation results comprehensively considered the human factors in human-machine interaction processes while reducing the influence of human factors during the evaluation process.The proposed method for evaluating digital human-machine interface in nuclear power plants was validated as effective for improving interface quality,reducing operators’ cognitive load and workload,and minimizing the risk of human errors,thereby proving the rationality of the evaluation method.The design and optimization of digital human-machine interface in nuclear power plants based on the evaluation results has potential practical application value.