A novel multilayer neural network model, called artificial synaptic network, was designed and implemented for single sensor localization with time-of-arrival (TOA) measurements. In the TOA localization problem, the location of a source sensor is estimated based on its distance from a number of anchor sensors. The measured distance values are noisy and the estimator should be able to handle different amounts of noise. Three neural network models: the proposed artificial synaptic network, a multi-layer perceptron network, and a generalized radial basis functions network were applied to the TOA localization problem. The performance of the models was compared with one another. The efficiency of the models was calculated based on the memory cost. The study result shows that the proposed artificial synaptic network has the lowest RMS error and highest efficiency. The robustness of the artificial synaptic network was compared with that of the least square (LS) method and the weighted least square (WLS) method. The Cramer-Rao lower bound (CRLB) of TOA localization was used as a benchmark. The model's robustness in high noise is better than the WLS method and remarkably close to the CRLB.

A multilayer feedforward neural network with two hidden layers was designed and developed for prediction of the phosphorus content of electroless Ni-P coatings. The input parameters of the network were the pH, metal turnover, and loading of an electroless bath. The output parameter was the phosphorus content of the electroless Ni-P coatings. The temperature and molar rate of the bath were constant (91 C; 0:4 Ni þþ =H 2 PO ÀÀ 2 ). The network was trained and tested using the data gathered from our own experiments. The goal of the study was to estimate the accuracy of this type of neural network in prediction of the phosphorus content. The study result shows that this type of network has high accuracy even when the number of hidden neurons is very low. Some comparison between the network's predictions and own experimental data are given.

A novel multilayer neural network was designed and implemented for prediction of the hardness of electroless Ni-P coatings. Heat treatment, a process for adjusting the hardness of electroless Ni-P coatings, was modeled. Three neural network models, a multilayer preceptron, a radial basis functions network, and a novel model, called the decomposercomposer model, were implemented and applied to the problem. The input parameters were the phosphorus content of the coatings, and the temperature and duration of the heat treatment process. The models output was the hardness of electroless Ni-P coatings. The training and test data were extracted from a number of experimental projects. The decomposer-composer model achieved better result and performance compared to the other models.