An artificial neural-network model for predicting the failure rate of De Havilland Dash-8 airplane tires utilizing the two-layered feedforward back-propagation algorithm as a learning rule is developed. The inputs to the neural network are independent variables, and the output is the failure rate of the tires. Six years of data are used for model building and validation. Model validation, which reflects the suitability of the model for future prediction, is performed by comparing the predictions of the model with that of the Weibull regression model. The results show that the failure rate predicted by the artificial neural network more closely agrees with the actual data than the failure rate predicted by the Weibull model.
Introduction
TIRES of airplanes, like tires of automobiles, are subjected to a number of wear-out processes, for example, uniform wear, accelerated wear at certain spots, microchipping, localized tire deformation, and so forth. In the case of airplanes, when the tires are in contact with the runway on landing, the conditions of wear are far more severe than the corresponding conditions in automobiles on highways. In the case of airplanes, the loads are not as uniform, there is a variety of shock loads, and a severe load spectrum is generated that can cause accelerated wear. Because tires are important aircraft components and the safety of an aircraft greatly depends on the reliability of its tires, their periodic monitoring and preventive maintenance are essential measures to increase aircraft reliability and are crucial for safe takeoff and landing. Tire life is defined by the wear limits set by controlling aviation agencies. When the tire damage due to wear-out processes reaches this critical limit, the tire is considered to have failed. The time to reach this critical manifestation of wear can be measured either by associated flight time or in terms of number of landings. It can also be written as t ∝ t r and t ∝ l where t is the flight operational time, t r is the time that the airplane tires are in contact with runway, and l is the number of landings. The tire life is not a fixed value but rather a random quantity, which is determined by t, bounded by t o < t < ∞, where t o is the minimum expected life.Accurately modeling the failure rate of airplane tires is of prime importance. This model should accurately predict the time of failure to avoid crashes during landing or takeoff. Various conventional regression models can be developed to model this failure rate. However, much interest has recently been focused on the application of artificial neural-network (ANN) modeling, 1−7 and it has been shown that the ANN performs better than the regression models.The ability of ANN to model multivariate problems without making complex dependency assumptions among the input variables is an advantage over the statistical models. Moreover, ANN extracts the implicit nonlinear relationships among the input variables through a learning process from the training data set. These features make neural netwo...
