No description available.

Accurate short term load forecasting plays a very important role in power system management. As electrical load data is highly non-linear in nature, in the proposed approach, we first separate out the linear and the non-linear parts, and then forecast the load using the non-linear part only. The Semiparametric spectral estimation method is used to decompose a load data signal into a harmonic linear signal model and a nonlinear trend. A support vector machine is then used to predict the non-linear trend. The final predicted signal is then found by adding the support vector machine predicted trend and the linear signal part. With careful determination of the linear component, the performance of the proposed method seems to be more robust than using only the raw load data, and in many cases the predicted signal of the proposed method is more accurate when we have only a small training set.

In financial econometrics the modeling of asset return series is closely related to the estimation of the corresponding conditional densities. One reason why one is interested in the whole conditional density and not only in the conditional mean, is that the conditional variance can be interpreted as a measure of time-dependent volatility of the return series. In fact, the modeling and the prediction of volatility is one of the central topics in asset pricing. In this paper we propose to estimate conditional densities semi-nonparametrically in a neural network framework. Our recurrent mixture density networks realize the basic ideas of prominent GARCH approaches but they are capable of modeling any continuous conditional density also allowing for time-dependent higher-order moments. Our empirical analysis on daily DAX data shows that out-of-sample volatility predictions of the neural network model are superior to predictions of GARCH models in that they have a higher correlation with implied volatilities. (author's abstract) / Series: Report Series SFB "Adaptive Information Systems and Modelling in Economics and Management Science"

This report examines the trends in sick leave usage among nurses in a hospital and aims at creating a forecasting model to predict sick leave usage on a weekly basis using the concept of artificial neural networks (ANN). The data used for the research includes the absenteeism (sick leave) reports for 3 years at a hospital. The analysis shows that there are certain factors that lead to a rise or fall in the weekly sick leave usage. The ANN model tries to capture the effect of these factors and forecasts the sick leave usage for a 1 year horizon based on what it has learned from the behavior of the historical data from the previous 2 years. The various parameters of the model are determined and the model is constructed and tested for its forecasting ability. / text

published_or_final_version / Applied Statistics / Master / Master of Social Sciences

published_or_final_version / Geography and Geology / Master / Master of Philosophy

published_or_final_version / Civil Engineering / Master / Master of Philosophy

The installed base of photovoltaic power plants in the United States has roughly doubled every 1 to 2 years between 2008 and 2015. The primary economic drivers of this are government mandates for renewable power, falling prices for all PV system components, 3rd party ownership models, and a generous tariff scheme known as net-metering. Other drivers include a desire for decreasing the environmental impact of electricity generation and a desire for some degree of independence from the local electric utility. The result is that in coming years, PV power will move from being a minor niche to a mainstream source of energy. As additional PV power comes online this will create challenges for the electric grid operators. We examine some problems related to large scale adoption of PV power in the United States. We do this by first discussing questions of reliability and efficiency at the PV system level. We measure the output of a fleet of small PV systems installed at Tucson Electric Power, and we characterize the degradation of those PV systems over several years. We develop methods to predict energy output from PV systems and quantify the impact of negatives such as partial shading, inverter inefficiency and malfunction of bypass diodes. Later we characterize the variability from large PV systems, including fleets of geographically diverse utility scale power plants. We also consider the power and energy requirements needed to smooth those systems, both from the perspective of an individual system and as a fleet. Finally we report on experiments from a utility scale PV plus battery hybrid system deployed near Tucson, Arizona where we characterize the ability of this system to produce smoothly ramping power as well as production of ancillary energy services such as frequency response.

No description available.

One of the most important aspects of coastal management and planning programs that needs to be investigated is shoreline dynamics. Long-term coastal analysis uses historical data to identify the sectors along the coast where the shoreline position has changed. Among the information that can be obtained from these studies are the general trend of coasts, either advancing or retreating. The erosion or accretion rates at each location can be used to forecast future shoreline positions. The current techniques used to study shoreline evolution are generally based on transects perpendicular to a baseline at selected points. But these techniques proved to be less efficient along more complex shorelines, and need to be refined. A new and more reliable method, the topologically constrained transect method (TCTM), was developed for this study and tested using data available for three sectors of the Texas Gulf Coast. Output data generated from TCTM also allowed performing shoreline evolution analysis and forecasting based on historical positions. Using topological constrained transects, this study provides a new method to estimate total areas of accretion or erosion at each segment of the coastline. Reliable estimates of future gains or losses of land along the coast will be extremely useful for planning and management decisions, especially those related to infrastructure and environmental impacts, and in the development of coastal models. Especially important is the potential to quickly identify areas of significant change, which eliminates the need for preliminary random sample surveying, and concentrate higher-resolution analyses in the most significant places. The results obtained in this research using the new methodology show that the Texas coast generally experiences erosion, with anthropogenic factors responsible for accretion. Accretion areas are located near coastal infrastructure, especially jetties that block the along shore sediment transport. The maximum erosion rate obtained in the study area is 5.48 m/year. This value helps make us aware of the powerful dynamic of the sector.