Research On Neural Network Prediction Based On Air China Flight Failure Coefficient

With the rapid development of civil aviation and the significant improvement of people's living standards,traveling by airplane has become an efficient way of travel,but due to the flight characteristics of aircraft and the precise characteristics of the aircraft structure,flight delays and flight accidents are also Repeated occurrences,and the life-threatening factor of the aircraft after a flight accident is also the highest among all kinds of vehicles.In recent years,various types of civil aviation flight accidents have occurred frequently,causing major economic losses and passenger casualties,eliminating aviation accidents caused by sporadic environmental factors.If a large number of flight data is used to effectively predict the failure rate of aircraft equipment,timely replacement The spare parts of related equipment will greatly reduce the delay,return and accident probability of the aircraft,improve flight reliability,and ensure the safety of life and property for passengers.Aircraft failures fall into many categories,engine failures,landing gear failures,door failures,control room equipment failures,etc.Most of the current research on aircraft failures focuses on engine failure studies,however due to the particularity of aircraft construction.Various types of faults may cause the aircraft to be dangerous during flight.Therefore,the main research objective of this paper is to establish a high-precision BP neural network prediction model to predict the occurrence of aircraft faults.The main idea is:in the case of massive flight data as the predicted sample data,due to the large number of data classification and complex latitude,in the preliminary processing method of data,the sample data is processed by the big data processing technology first,and the sample data is classified.To reduce the complexity of the data category,and reduce the computational cost of the neural network model in the hidden layer;due to the inherent defects of the BP neural network,the local search process in the hidden layer tends to fall into the local minimum.Therefore,it is compensated by the SA algorithm.In the search optimization process,the global search is optimal,and the local optimum is jumped out,so that the local optimal defects that are easy to appear in the BP algorithm are optimized.The result of the BP algorithm also depends on the initial weight and Threshold,the initial weights and thresholds are optimized by the GA genetic algorithm,so that the optimized BP neural network can have a more accurate prediction function output.The research goal of this paper is to predict the occurrence of aircraft faults based on big data processing and neural network prediction model.It can be seen from the experimental results that the results are compared with the results of a single BP neural network model prediction algorithm.By optimizing the BP neural network by GA genetic algorithm and SA-GA simulated annealing genetic algorithm,the improved algorithm not only improves the prediction accuracy of the model,but also makes up for the shortcomings of the BP neural network algorithm.Training takes up time and achieves the desired predictions.In the later stage,the calculation process and data selection of the algorithm can be further optimized,and still have a high potential for improvement.