Complex Artificial Neural Networks Research Articles

Information distribution on regions of speckle patterns for imaging of multimode fiber

Multimode fibers (MMF) have been extensively investigated for transmitting images. The transmitting images are distorted into speckle patterns by MMFs, which can be reconstructed by neural networks. We studied the information distribution of MMF speckle patterns for image reconstruction. The speckle patterns, segmented by three methods of segmentation, as Centering (1), Quartering (2) and Surrounding (3), are reconstructed into input images by Complex Artificial Neural Network (CANN). Experimental results show that only about one third of full speckle patterns is enough to reconstruct the original images. The quality of reconstructed image is related to the cropping method with different frequency components in speckle patterns, under the same cropped size, Centering segmentation has 4% performance improvement compared to Surrounding segmentation. Optimized segmentation will improve the quality of reconstructed images.

Определение упругих и диссипативных свойств материалов с помощью сочетания метода конечных элементов и комплекснозначных искусственных нейронных сетей

The application of the complex artificial neural networks (CANN) to the inverse identification problem of the elastic and dissipative properties of deformable solids is considered. The additional information to the inverse problem is components of the displacement vector measured in a set of points at the solid boundary (positional measurement). This solid performs harmonic oscillations in the first resonant frequency. The process of displacement measurement is simulated using the calculation of finite elements software ANSYS, the building of the amplitude-frequency characteristics (AFC) of the displacement, and of the selection of their values for a set of frequencies (frequency measurement). In the given numerical example, problems on the accurate identification of the elastic modulus, and material quality depending on the number of measure points and their location, as well as on the neural network architecture and the length of the training process performed by the complex-value error back propagation (CBP) algorithm are investigated.

Bad data analysis in power system measurement estimation using complex artificial neural network based on the extended complex Kalman filter

AbstractThis paper proposes a method for bad data analysis in power system measurement estimation using complex artificial neural network (CANN) based on the extended complex Kalman filter (ECKF). The proposed algorithm is better in noise immunity since the link weighting in the CANN can be automatically adjusted with trained data through the ECKF. Moreover, the CANN is quite suitable for complex training data such as complex power in a power system since its input and output performs a nonlinear mapping. Four systems including a 6‐bus system, the IEEE 30‐bus system, IEEE 118‐bus system, and a practical system are used as examples to verify the feasibility of the ECKF‐CANN approach. Results show the proposed algorithm has increased sensitivity in identifying gross measurement errors with respect to the standard ANN. Copyright © 2009 John Wiley & Sons, Ltd.