Air traffic controllers workload forecasting method based on neural network

Using the routinely-recorded flight data, the values of 10 air traffic complexity evaluation metrics are calculated, and the BP neutral network model for air traffic controller workload prediction is formed and verified through pilot-controller voice communication data. The number of hidden layer nodes were screened using the trial and error method. When the topological structure of BP neural network is 10-4-1, a smaller level of network training error is achieved, with less iterations and training time. By testing the network with the collected pilot-controller voice communication data, the predicted data are found to be in relatively good consistency with the collected ones, which indicates that BP network is an effective method to predict air traffic controller workload.