Performance of neural network based restoration of optical networks is evaluated and a few possible improvements are proposed. Neural network based restoration is simulated with optical link capacities assigned by a new method. Two new improvement methods are developed to reduce the neural network size and the restoration time of severed optical connections. Cycle based restoration is suggested, which reduces the neural network structure by restoring the severed connections for each optical node, iteratively. Additionally, to reduce the restoration time, the necessary waiting time before the neuron outputs fire for the propagation delay over the network is computed and embedded in the control structure of the neural network. The improvement methods are evaluated by simulations in terms of restorability, restoration time, network redundancy and average length of restoration paths for different failure cases and different security requirements.

Lakes present in arid regions of Central Anatolia need further attention with regard to water quality. In most cases, mathematical modeling is a helpful tool that might be used to predict the DO concentration of a lake. Deterministic models are frequently used to describe the system behavior. However most ecological systems are so complex and unstable. In case, the deterministic models have high chance of failure due to absence of priori information. For such cases black box models might be essential. In this study DO in Eymir Lake located in Ankara was modeled by using both Artificial Neural Networks (ANN) and Adaptive Neuro Fuzzy Inference System (ANFIS). Phosphate, Orthophospate, pH, Chlorophyll-a, Temperature, Alkalinity, Nitrate, Total Kjeldahl Nitrogen, Wind, Precipitation, Air Temperature were the input parameters of ANN and ANFIS. The aims of these modeling studies were: to develop models with ANN to predict DO concentration in Lake Eymir with high fidelity to actual DO data, to compare the success (prediction capacity) of ANN and ANFIS on DO modeling, to determine the degree of dependence of different parameters on DO. For modeling studies &ldquo / Matlab R 2007b&rdquo / software was used. The results indicated that ANN has high prediction capacity of DO and ANFIS has low with respect to ANN. Failure of ANFIS was due to low functionality of Matlab ANFIS Graphical User Interface. For ANN Modeling effect of meteorological data on DO data on surface of the lake was successfully described and summer month super saturation DO concentrations were successfully predicted.

The main objective of this work is to develop models using multivariable regression and artificial neural network approaches for cost estimation of the construction costs of trackworks of light rail transit and metro projects at the early stages of the construction process in Turkey. These two approaches were applied to a data set of 16 projects by using seventeen parameters available at the early design phase. According to the results of each method, regression analysis estimated the cost of testing samples with an error of 2.32%. On the other hand, artificial neural network estimated the cost with 5.76% error, which is slightly higher than the regression error. As a result, two successful cost estimation models have been developed within the scope of this study. These models can be beneficial while taking the decision in the tender phase of projects that includes trackworks.

Designing smarter vehicles, aiming to minimize the number of driverbased wrong decisions or accidents, which can be faced with during the drive, is one of hot topics of today&rsquo / s automotive technology. In the design of smarter vehicles, several research issues can be addressed / one of which is Traffic Sign Recognition (TSR). In TSR systems, the aim is to remind or warn drivers about the restrictions, dangers or other information imparted by traffic signs, beforehand. Since the existing signs are designed to draw drivers&rsquo / attention by their colors and shapes, processing of these features is one of the crucial parts in these systems. In this thesis, a Traffic Sign Recognition System, having ability of detection and identification of traffic signs even with bad visual artifacts those originate from some weather conditions or other circumstances, is developed. The developed algorithm in this thesis, segments the required color influenced by the illumination of the environment, then reconstructs the shape of partially occluded traffic sign by its remaining segments and finally, identifies it. These three stages are called as &ldquo / Segmentation&rdquo / , &ldquo / Reconstruction&rdquo / and &ldquo / Identification&rdquo / respectively, within this thesis. Due to the difficulty of analyzing partial segments to construct the main frame (a whole sign), the main complexity of the algorithm takes place in the &ldquo / Reconstruction&rdquo / stage.

Chlorophyll-a is a parameter which can be used to understand the trophic state of water bodies. Therefore, monitoring of this parameter is required. Yet, distribution of chlorophyll-a in water bodies is not homogeneous and exhibits both spatial and temporal variations. Therefore, frequent sampling and high sample sizes are needed for the determination of chlorophyll-a quantities. This would in return increase the sampling costs and labor requirement, especially if the topography makes the location hard to reach. Remote sensing is a technology that can aid in handling of these difficulties and obtain a continuous distribution of chlorophyll-a concentrations in a water body. In this method, reflectance from water bodies in different wavelengths is used to quantify the chlorophyll-a concentrations. In previous studies in literature, empirical regression models that use the reflectance values in different bands in different combinations have been derived. Yet, prediction performances of these models decline especially in shallow lakes. In this study, the spatial distribution of chlorophyll-a in shallow Lake Eymir is determined using both regression models and artificial neural network models that use hybrid inputs. Unlike the models generated before, field measured parameters which can influence the reflectance values in remotely sensed images have been used in addition to the reflectance values. The parameters that are considered other than reflectance values are photosynthetically active radiation (PAR), secchi depth (SD), water column depth, turbidity, dissolved oxygen concentration (DO), pH, total suspended solids (TSS), total dissolved organic matter (TDOM), water and air temperatures, wind data and humidity. Reflectance values are obtained from QuickBird and World View 2 satellite images. Effect of using hybrid input in mapping the reflectance values to chlorophyll-a concentrations are studied. In the context of this study, three different high-resolution satellite images are analyzed for the spatial distribution of chlorophyll-a concentration in Lake Eymir. Field and laboratory studies are conducted for the measurement of parameters other than the reflectance values. Principle component analysis is applied on the collected data to decrease the number of model input parameters. Then, linear and non-linear regression and artificial neural network (ANN) models are derived to model the chlorophyll-a concentrations in Lake Eymir. Results indicate that ANN model shows better predictability compared to regression models. The predictability of ANN model increases with increasing variation in the dataset. Finally, it is seen that in determination of chlorophyll-a concentrations using remotely sensed data, models with hybrid inputs are superior compared to ones that use only remotely sensed reflectance values.

A risk assessment model (MAcRisk) have been developed to forecast the probability and the risk of maritime accidents on Bosphorus. Accident archives of Undersecretariat Maritime Affairs Search and Rescue Department, weather conditions data of Turkish State Meteorological Service and bathymetry and current maps of Office of Navigation, Hydrography and Oceanography have been used to prepare the model input and to forecast the accident probability. Accident data has been compiled according to stated sub-regions on Bosphorus and event type of accidents such as collision, grounding, capsizing, fire and other. All data that could be obtained are used to clarify the relationship on accident reasons. An artificial neural network model has been developed to forecast the maritime accidents in Bosphorus.

Conceptual estimation techniques play an important role in determining the approximate costs of construction projects especially during feasibility stages. Moreover, pre-design estimates are also crucial for the contractors. With the help of the conceptual predictions companies can determine approximate project costs and can gain several advantages before tendering phase. The main objective of this thesis is to focus on modeling of quantities instead of costs and to develop quantity take-off models for pre-design cost estimation of bridge projects. Majority of the existing studies focus on modeling of costs for conceptual cost estimation. This study includes modeling of the quantity take off items in a specific single span highway bridge using three different techniques namely, linear regression, neural network and case based reasoning. During this study 40 single span highway bridge projects whose owner is Republic of Turkey General Directorate of Railways, Ports and Airports Constructions were investigated and models for each work item were developed. Then by integrating the quantity take off estimations with unit costs, total project costs were calculated. As a result by evaluating the prediction performance of the models, comparison of the methods was achieved. Results are discussed along with the advantages of the proposed method for conceptual cost estimation of bridge projects.

Along with advancement of human industrialization, pollution in the ocean is getting worse. Moreover, the overfishing through the years has caused catastrophic damage to the ocean eco-system. In order to avoid exhaustion of fishery resource, many concepts of planned administrative fishery has become popular, and thereamong, ocean ranch draws the most attention. Artificial reef plays a key role in an ocean ranch, which starts with incubating brood fish in the laboratory. Often, the brood fish will grow in the cage near coast till proper size, then be released to the artificial reef. If fish groups do not disperse and multiply, the artificial reef can be considered successful. The success of the artificial reef relies on the stable foundation. Consequently, the composition of seabed sediment under the planned site should be investigated thoroughly before hand. This research introduced a remote investigation method, which an active sonar, depth sounder, was used to emit and collect acoustic signals. By using the signals reflected from the seabed, the sediment composition can be analyzed. However, all acoustic signals are subjected to noise through propagation, and distorted somehow. Therefore, certain signal pre-processing should be applied to the received signal, and representative characteristics can be extracted from it. In this research, the recognition platform was built on artificial neural network (ANN) in this research. Among many network algorithm modes, this research chose the widely used backpropagation learning algorithm to be the main structure in ANN. The goal of this research was to discriminate among three seabed sediments: fine sand, medium sand, and rock. During the signal processing, characteristics were extracted by using peak value selection method. Selected major frequency peaks were fed into the network to train and learn. According to partial error relation between recognition and practical result, weights of the network were adjusted for improving successful ratio. Finally, a reliable acoustic wave signal recognition system was constructed.

None

A burn injury is a disastrous trauma and can have very wide ranging impacts, including individual, family, and social. Burns patients generally have a long period of hospital stay whose accurate prediction can not only facilitate allocations of scarce medical resources but also help clinicians to counsel patients and relatives at an early stage of care. Besides prediction accuracy, prediction timing of length of hospital stay (LOS) for burn patients is also critical. Early prediction has profound effects on more efficient and effective medical resource allocations and better patient care and management. Hence, the objective of this study is to apply a backpropagation neural network (BPNN) for predicting length of hospital stay (LOS) for burn patients at early stages of care. Specifically, we defined two early-prediction timing, including admission and initial treatment stages. Prediction timing at the admission stage is to predict a burn patient¡¦s LOS when the patient is admitted into the Burns Unit. Prediction at the initial treatment stage refers to the timing right after the first surgery for burn wound excision and skin graft is performed (typically within 72 hours of injury if the patient¡¦s condition allows). Experimentally, we evaluated the prediction accuracy of these two stages, using that achieved at the post-treatment stage (referring to the timing when all surgeries for burn wound excision and skin graft are performed) as benchmarks. The evaluation results showed that prediction LOS at the admission and the initial treatment stages could attain an accuracy of 50.37% and 57.22%, respectively. Compared to the accuracy of 62.13% achieved by the post-treatment stage, the performance reached by the initial treatment stage would consider satisfactory.