Research On Assessment Method Of Min Crew Workload Application To Civil Aircraft Cockpit

Civil aviators often process lots of information and make rapid responses and decisionsaccording to it in a short time in Air Bus cockpit. During the flight, if the workload does not match thenumber of the pilots,the attention of the pilots may be not centralized or loaf on the job. The situationeven can cause serious aviation accidents,like air crash.Therefore,how to evaluate the crewworkload and determine the number of the crew appear to be particularly important.In this paper,theevaluation methods of the crew workload are deeply studied, and simulated flight experiment andevaluation model are designed to support the evaluation of crew workload.Combining with the experiment platform, testees, procedures, experimental data collection andprocessing, the flight simulation experiment is designed basing on flight procedures to collect thephysiological indicators and the measurement of the workload.During the experiment,the crewworkload is distinguished and the physiological indicators are changed by setting different failuremodes and weather conditions and record the data at the same time. The relativity betweenphysiological indicators and NASA-TLX is proved by the SPSS.Affected by many factors and complicated evaluation process are characteristics of crewworkload. To solve this problem, a crew workload evaluation model is built basing on BP neuralnetwork because of its approaching any nonlinear function ability and Fault-tolerant ability.In thisevaluation model, physiological indicators are as input and the value of NASA-TLX as output. Easilyfalling into local optimum is the defect of BP neural network.To solve this problem, GABP model isbuilt which is based on GA.BP and GABP evaluation models are realized by MATLAB. Train the two models by using thedata from the experiment and predict the crew workload. The two evaluation models are proved to befeasible by comparing the predicted values and the expectations.GA is proved to be effective inoptimizing the predict precision of crew workload evaluation model by comparing the generalizationabilities of the two models.