Development of probabilistic constant life diagrams using modular networks

Abstract

This study consists in the evaluation of the use of an artificial neural network of modular architecture in building probabilistic constant life diagrams. Therefore, an algorithm developed in previous studies which was applied to achieve deterministic values has proved itself viable when at least three S-N curves were used. For this case, the probability S-N curves were used for training and validation of the modular network based on the generalized power law and a probability of 5% for failure has been considered. In addition, three composite materials were evaluated with a considerable number of tests to better assess the model.

Fatigue modeling using neural networks: A comprehensive review

“…Generating data for ANN training, based on prior knowledge of the problem (trivial data such as the average stress and number of cycles when the alternating stress is zero) to train the ANN is an interesting solution. This is done in previous works 74,76–78,91,107,108 …”

“…Next, the fatigue life, its corresponding failure probability, and other variables such as the stress level are used to train the NN. This methodology is used in work conducted by Barbosa et al, 106 Lee et al, 8 Belísio et al, 107 and da Cunha Diniz and Júnior 108…”

Using probabilistic neural networks for modeling metal fatigue and random vibration in process pipework

“…In this section, we demonstrate the advantages of PNN in comparison with ANN. First, we introduce the data that was used in Belísio et al 22 in Sections 4.1 and 4.2. Then we demonstrate in Section 4.3 the gains that can be achieved in the performance and stability of the network when the P50% and P5% networks are combined in the same network.…”

“…The results showed that ANNs have clear advantages in developing S‐N curves in the CLD model using only a small number of data points produced experimentally or numerically. To deal with the inherited uncertainty, Belísio et al 22 developed a new method to generate probabilistic CLD models using only the Weibull distribution. Barbosa et al 23 built a probabilistic CLD (Haigh diagram) using ANNs where the uncertainty was accounted for using probabilistic Stüssi fatigue S‐N fields and Weibull distribution.…”

Estimating fatigue behavior of a family of aluminum overhead conductors using ANNs

“…To obtain additional details on these fatigue tests and on the conductors specifications interested readers should refer to Fadel, 13 Pestana, 14 The multilayer perceptron (trained with the backpropagation algorithm 16 ) was used to develop the ANN. 6 This model, which has been applied successfully to resolve many engineering problems, [17][18][19][20][21][22][23][24][25][26] can simulate material/component behavior by means of ANN training.…”