An entropy weight variable fuzzy recognition and 1D-CNN deep learning method for general aviation fleet reliability evaluation

Abstract

The reliability level of general aviation fleet system directly affects the economic benefits and safe operation of general aviation fleet. In order to effectively evaluate the reliability level of general aviation fleet, using the entropy weight variable fuzzy recognition and 1D-CNN depth learning reliability evaluation method. Firstly, taking the Cessna 172 general aviation fleet as the research object, refers to the maintenance statistical analysis of general aviation fleet reliability data, and classifies the fleet reliability evaluation indexes according to the ATA100 chapter standard. Combined with index importance analysis and Delphi expert investigation, 14 key items are extracted as reliability evaluation indexes of general aviation fleet. Secondly, using entropy weight method to obtain indexes weight objectively, and the evaluation level membership function is constructed based on variable fuzzy recognition method. Finally, a reliability evaluation model based on 1D-CNN deep learning method was established. Through training and testing the reliability data evaluation model of general aviation fleet, and comparing with the results of evaluation methods such as support vector machines. The results show that the recognition rate of the 1D-CNN deep learning method based on entropy weight variable fuzzy recognition can reach 91.95%, verifying the objective effectiveness of the evaluation method.