The School of Electrical, Electronic and Computer Engineering at the North-West University in Potchefstroom has established an active magnetic bearing (AMB) research group called McTronX. This group provides extensive knowledge and experience in the theory and application of AMBs. By making use of the expertise contained within McTronX and the rest of the control engineering community, an adaptive controller for an AMB flywheel system is implemented. The adaptive controller is faced with many challenges because AMB systems are multivariable, nonlinear, dynamic and inherently unstable systems. It is no wonder that existing AMB models are poor approximations of reality. This modelling problem is avoided because the adaptive controller is based on an indirect adaptive control law. Online system identification is performed by a neural network to obtain a better model of the AMB flywheel system. More specifically, a nonlinear autoregressive with exogenous inputs (NARX) neural network is implemented as an online observer. Changes in the AMB flywheel system’s environment also add uncertainty to the control problem. The adaptive controller adjusts to these changes as opposed to a robust controller which operates despite the changes. Making use of reinforcement learning because no online training data can be obtained, an adaptive critic model is applied. The adaptive controller consists of three neural networks: a critic, an actor and an observer. It is called an observer-based adaptive critic neural controller (ACNC). Genetic algorithms are used as global optimization tools to obtain values for the parameters of the observer, critic and actor. These parameters include the number of neurons and the learning rate for each neural network. Since the observer uses a different error signal than the actor and critic, its parameters are optimized separately. When the actor and critic parameters are optimized by minimizing the tracking error, the observer parameters are kept constant. The chosen adaptive control design boasts analytical proofs of stability using Lyapunov stability analysis methods. These proofs clearly confirm that the design ensures stable simultaneous identification and tracking of the AMB flywheel system. Performance verification is achieved by step response, robustness and stability analysis. The final adaptive control system remains stable in the presence of severe cross-coupling effects whereas the original decentralized PD control system destabilizes. This study provides the justification for further research into adaptive control using artificial intelligence techniques as applied to the AMB flywheel system. / Thesis (M.Ing. (Computer and Electronical Engineering))--North-West University, Potchefstroom Campus, 2011.

The statistical properties of training, validation and test data play an important role in assuring optimal performance in artificial neural networks (ANN). Re-searchers have proposed randomized data partitioning (RDP) and stratified data partitioning (SDP) methods for partition of input data into training, vali-dation and test datasets. RDP methods based on genetic algorithm (GA) are computationally expensive as the random search space can be in the power of twenty or more for an average sized dataset. For SDP methods, clustering al-gorithms such as self organizing map (SOM) and fuzzy clustering (FC) are used to form strata. It is assumed that data points in any individual stratum are in close statistical agreement. Reported clustering algorithms are designed to form natural clusters. In the case of large multivariate datasets, some of these natural clusters can be big enough such that the furthest data vectors are statis-tically far away from the mean. Further, these algorithms are computationally expensive as well. Here a custom design clustering algorithm (CDCA) has been proposed to overcome these shortcomings. Comparisons have been made using three benchmark case studies, one each from classification, function ap-proximation and prediction domain respectively. The proposed CDCA data partitioning method was evaluated in comparison with SOM, FC and GA based data partitioning methods. It was found that the CDCA data partitioning method not only performed well but also reduced the average CPU time. / Engineering Management

Footings are often founded on multi-layered soil profiles. Real soil profiles are often multi-layered with material constantly varying with depth, which affects the footing response significantly. Furthermore, the properties of the soil are known to vary with location. The spatial variability of soil can be described by random field theory and geostatistics. The research presented in this thesis focuses on quantifying the effect of soil variability on the bearing capacity of rough strip footings on single and two layered, purely-cohesive, spatially variable soil profiles. This has been achieved by using Monte Carlo analysis, where the rough strip footings are founded on simulated soil profiles are analysed using finite element limit analysis. The simulations of virtual soil profiles are carried out using Local Average Subdivision (LAS), a numerical model based on the random field theory. An extensive parametric study has been carried out and the results of the analyses are presented as normalized means and coefficients of variation of bearing capacity factor, and comparisons between different cases are presented. The results indicate that, in general, the mean of the bearing capacity reduces as soil variability increases and the worst case scenario occurs when the correlation length is in the range of 0.5 to 1.0 times the footing width. The problem of estimating the bearing capacity of shallow strip footings founded on multi-layered soil profiles is very complex, due to the incomplete knowledge of interactions and relationships between parameters. Much research has been carried out on single- and two-layered homogeneous soil profiles. At present, the inaccurate weighted average method is the only technique available for estimating the bearing capacity of footing on soils with three or more layers. In this research, artificial neural networks (ANNs) are used to develop meta-models for bearing capacity estimation. ANNs are numerical modelling techniques that imitate the human brain capability to learn from experience. This research is limited to shallow strip footing founded on soil mass consisting of ten layers, which are weightless, purely cohesive and cohesive-frictional. A large number of data has been obtained by using finite element limit analysis. These data are used to train and verify the ANN models. The shear strength (cohesion and friction angle), soil thickness, and footing width are used as model inputs, as they are influencing factors of bearing capacity of footings. The model outputs are the bearing capacities of the footings. The developed ANN-based models are then compared with the weighted average method. Hand-calculation design formulae for estimation of bearing capacity of footings on ten-layered soil profiles, based on the ANN models, are presented. It is shown that the ANN-based models have the ability to predict the bearing capacity of footings on ten-layered soil profiles with a high degree of accuracy, and outperform traditional methods. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1368281 / Thesis (Ph.D.) - University of Adelaide, School of Civil, Environmental and Mining Engineering, 2009

Internal Symmetry Networks are a recently developed class of Cellular Neural Network inspired by the phenomenon of internal symmetry in quantum physics. Their hidden unit activations are acted on non-trivially by the dihedral group of symmetries of the square. Here, we extend Internal Symmetry Networks to include recurrent connections, and train them by backpropagation to perform a variety of image processing tasks, smoothing, sharpening, edge detection, synthetic image segmentation, texture segmentation and object recognition. By a large number of experiments, we find some guidelines to construct appropriate configurations of the net for different tasks.

Footings are often founded on multi-layered soil profiles. Real soil profiles are often multi-layered with material constantly varying with depth, which affects the footing response significantly. Furthermore, the properties of the soil are known to vary with location. The spatial variability of soil can be described by random field theory and geostatistics. The research presented in this thesis focuses on quantifying the effect of soil variability on the bearing capacity of rough strip footings on single and two layered, purely-cohesive, spatially variable soil profiles. This has been achieved by using Monte Carlo analysis, where the rough strip footings are founded on simulated soil profiles are analysed using finite element limit analysis. The simulations of virtual soil profiles are carried out using Local Average Subdivision (LAS), a numerical model based on the random field theory. An extensive parametric study has been carried out and the results of the analyses are presented as normalized means and coefficients of variation of bearing capacity factor, and comparisons between different cases are presented. The results indicate that, in general, the mean of the bearing capacity reduces as soil variability increases and the worst case scenario occurs when the correlation length is in the range of 0.5 to 1.0 times the footing width. The problem of estimating the bearing capacity of shallow strip footings founded on multi-layered soil profiles is very complex, due to the incomplete knowledge of interactions and relationships between parameters. Much research has been carried out on single- and two-layered homogeneous soil profiles. At present, the inaccurate weighted average method is the only technique available for estimating the bearing capacity of footing on soils with three or more layers. In this research, artificial neural networks (ANNs) are used to develop meta-models for bearing capacity estimation. ANNs are numerical modelling techniques that imitate the human brain capability to learn from experience. This research is limited to shallow strip footing founded on soil mass consisting of ten layers, which are weightless, purely cohesive and cohesive-frictional. A large number of data has been obtained by using finite element limit analysis. These data are used to train and verify the ANN models. The shear strength (cohesion and friction angle), soil thickness, and footing width are used as model inputs, as they are influencing factors of bearing capacity of footings. The model outputs are the bearing capacities of the footings. The developed ANN-based models are then compared with the weighted average method. Hand-calculation design formulae for estimation of bearing capacity of footings on ten-layered soil profiles, based on the ANN models, are presented. It is shown that the ANN-based models have the ability to predict the bearing capacity of footings on ten-layered soil profiles with a high degree of accuracy, and outperform traditional methods. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1368281 / Thesis (Ph.D.) - University of Adelaide, School of Civil, Environmental and Mining Engineering, 2009

Internal Symmetry Networks are a recently developed class of Cellular Neural Network inspired by the phenomenon of internal symmetry in quantum physics. Their hidden unit activations are acted on non-trivially by the dihedral group of symmetries of the square. Here, we extend Internal Symmetry Networks to include recurrent connections, and train them by backpropagation to perform a variety of image processing tasks, smoothing, sharpening, edge detection, synthetic image segmentation, texture segmentation and object recognition. By a large number of experiments, we find some guidelines to construct appropriate configurations of the net for different tasks.

“Winner-take-all” networks typically pick as winners that alternative with the largest excitatory input. This choice is far from optimal when there is uncertainty in the strength of the inputs, and when information is available about how alternatives may be related. In the Social Choice community, many other procedures will yield more robust winners. The Borda Count and the pair-wise Condorcet tally are among the most favored. Their implementations are simple modifications of classical recurrent networks.

Broken fences is a big security risk for any facility or area with strict security standards. In this report we suggest a machine learning approach to automate the surveillance for chain-linked fences. The main challenge is to classify broken and non-broken fences with the help of a convolution neural network. Gathering data for this task is done by hand and the dataset is about 127 videos at 26 minutes length total on 23 different locations. The model and dataset are tested on three performances traits, scaling, augmentation improvement and false rate. In these tests we concluded that nearest neighbor increased accuracy. Classifying with fences that has been included in the training data a false rate that was low, about 1%. Classifying with fences that are unknown to the model produced a false rate of about 90%. With these results we concludes that this method and dataset is useful under the right circumstances but not in an unknown environment.

Object recognition systems have significant influences on modern life. Face, iris and finger point recognition applications are commonly applied for the security purposes; ASR (Automatic Speech Recognition) is commonly implemented on speech subtitle generation for various videos and audios, such as YouTube; HWR (Handwriting Recognition) systems are essential on the post office for cheque and postcode detection; ADAS (Advanced Driver Assistance System) are well applied to improve drivers’, passages’ and pedestrians’ safety. Object recognition techniques are crucial and valuable for academia, commerce and industry. Accuracy and efficiency are two important standards to evaluate the performance of recognition techniques. Accuracy includes how many objects can be indicated in real scene and how many of them can be correctly classified. Efficiency means speed for system training and sample testing. Traditional object detecting methods, such as HOG (Histogram of orientated Gradient) feature detector combining with SVM (Support Vector Machine) classifier, cannot compete with frameworks of neural networks in both efficiency and accuracy. Since neural network has better performance and potential for improvement, it is worth to gain insight into this field to design more advanced recognition systems. In this thesis, we list and analyze sophisticated techniques and frameworks for object recognition. To understand the mathematical theory for network design, state-of-the-art networks in ILSVRC (ImageNET Large Scale Visual Recognition Challenge) are studied. Based on analysis and the concept of edge detectors, a simple CNN (Convolutional Neural Network) structure is designed as a trail to explore the possibility to utilize the network of high width and low depth for region proposal selection, object recognition and target region refining. We adopt Le-Net as the template, taking advantage of multi-kernels of GoogLe-Net. We made experiments to test the performance of this simple structure of the vehicle and face through ImageNet dataset. The accuracy for the single object detection is 81% in average and for plural object detection is 73.5%. We refined networks through many aspects to reach the final accuracy 95% for single object detection and 89% for plural object detection.

In this thesis, the neural network, a powerful tool which has demonstrated its ability in many fields, is studied for the indoor localization system and channel prediction system. This thesis first proposes a received signal strength indicator (RSSI) fingerprinting-based indoor positioning system for the widely deployed WiFi environment, using deep neural networks (DNN). To reduce the computing time as well as improve the estimation accuracy, a two-step scheme is designed, employing a classification network for clustering and several regression networks for final location prediction. A new fingerprinting, which utilizes the similarity in RSSI readings of the nearby reference points (RPs) is also proposed. Real-time tests demonstrate that the proposed algorithm achieves an average distance error of 43.5 inches. Then this thesis extends the ability of the neural network to the physical layer communications by introducing a recurrent neural network (RNN) based approach for real-time channel prediction which uses the recent history channel state information (CSI) estimation for online training before prediction, to adapt to the continuously changing channel to gain a more accurate CSI prediction compared to the other conventional methods. Furthermore, the proposed method needs no additional knowledge, neither the internal properties of the channel itself nor the external features that affect the channel propagation. The proposed approach outperforms the other methods in a changing environment in the simulation test, validating it a promising method for channel prediction in wireless communications. / Graduate