224 research outputs found
Artificial Neural Network with Adaptive Multidimensional Spline Activation Functions
No full textThis work concerns a new kind of neural structure that involves a multidimensional adaptive activation function. The proposed architecture, based on multidimensional cubic spline, allows to collect information from the previous network layer in aggregate form. In other words the number of network connections (structural complexity) can be very low respect to the problem complexity. This fact, as experimentally demonstrated in the paper, improve the network generalization capabilities and speed up the convergence of the learning process. A specific learning algorithm is derived and experimental results demonstrate the effectiveness of the proposed architecture
Ship to shore crane subject to earthquake
No full textThis paper is concerned the dynamical behaviour of the lifting machine and in particular the ship to shore crane subject to the earthquake. The reason of this work starts at the observation that the components of the crane may be subjected to very high stress and so the earthquake may be very dangerous. There are many phenomena induced by the earthquake like local and global buckling, and overturning moment which may lead to the collapse of the crane components. The paper wants to answer to the question if the earthquake is more dangerous than the normal operating conditions for the ship to shore crane and if the answer is yes, it is very important to know the differences. For this scope, the first step regards the whole design, by the standards, a very big ship to shore crane. The main parameters are: • height=33+15 m; • working load limit=450 kN.
After that, a solid model and subsequent fem analyses were performed in order to verify the safety factors for different load conditions that the crane may be subjected. The next step regards the dynamical analyses like modal analyses for evaluation both the deformation and natural frequencies and both the modal participation factors. The last step concerns the application of specific spectrums of acceleration to simulate a possible earthquake in the Italian region. The results shows that, in this specific case , the actions induce by earthquake are very low
Experimental and analytical study of random fatigue, in time and frequencies domain, on an industrial wheel
No full textIndustrial wheels are components subject to fatigue under a high number of stress cycles, depending on the type of vehicle they are installed to. The reliability of these components is strictly dependent on the accuracy of the fatigue validation method adopted. A common practice in the industry is to test loads under constant amplitude in laboratory through fatigue and test parameters according to the standards used in the industry. The object of this research is to compare the results in terms of damage and the number of failure cycles, adopting both the time domain and frequency domain approach on a real industrial component. Different theories were developed and applied to the real case of study of an industrial wheel, under specific load cases. Eventually, we applied different criteria to the numerical analysis. The added value of this study is the application of the fatigue criteria on a real industrial component, in order to check the reliability of the criteria. Eventually, we determined that the results show that, for these specific experimental load conditions, frequency-domain methods are a bit more conservative than time-domain methods
Adaptive Multidimensional Spline Neural Network for Digital Equalization
No full textPresents a new neural architecture that is suitable for digital signal processing applications. The architecture, which is based on adaptable multidimensional activation functions, allows one to collect information from the previous network layer in aggregate form. In other words, the number of network connections (the structural complexity) can be very low with respect to the problem complexity. This fact, as experimentally demonstrated in this paper, improves the network's generalization capabilities and speeds up the convergence of the learning process. A specific learning algorithm is derived, and experimental results on channel equalization demonstrate the effectiveness of the proposed architecture
Comparative reliability analyses of pressure vessel made of steel, aluminum alloy and composite material
No full textReliability design of a pressure vessel made of composite materials
No full textThe focus of this research is to evaluate the reliability of pressure vessels made of composite materials (carbon and glass fiber) with respect to stress and burst pressure. Small thicknesses theory was adopted to sizing the components. The geometry was verified through FEM analyses. To compare the results, also pressure vessel, made of steel and aluminum alloy, were studied. The reliability of the components and the deterministic safety coefficients were evaluated considering the variability of geometric characteristics, the materials properties and the number of layers wrapped. From the results, it emerges that the weight of pressure vessel made of carbon fibers is equal to 17% to the one made of steel, that the relationship between deterministic coefficient and reliability is highly non-linear and the reliability of the components made of composite materials, are similar to those made of steel or aluminum alloy
Spline neural networks for blind separation of post-nonlinear-linear mixtures
No full textIn this paper, a novel paradigm for blind source separation in the presence of nonlinear mixtures is presented. In particular the paper addresses the problem of post-nonlinear mixing followed by another instantaneous mixing system. This model is called here the post-nonlinear-linear model. The method is based on the use of the recently introduced flexible. activation function whose control points are adaptively changed: a neural model based on adaptive B-spline functions is employed. The signal separation is achieved through-an information maximization criterion. Experimental results and comparison with existing solutions confirm the effectiveness of the proposed architecture
- …
