Rafael C. Leme

A multivariate robust parameter optimization approach based on Principal Component Analysis with combined arrays

Todays modern industries have found a wide array of applications for optimization methods based on modeling with Robust Parameter Designs (RPD). Methods of carrying out RPD have thus multiplied. However, little attention has been given to the multiobjective optimization of correlated multiple responses using response surface with combined arrays. Considering this gap, this paper presents a multiobjective hybrid approach combining response surface methodology (RSM) with Principal Component Analysis (PCA) to study a multi-response dataset with an embedded noise factor, using a DOE combined array. How this approach differs from the most common approaches to RPD is that it derives the mean and variance equations using the propagation of error principle (POE). This comes from a control-noise response surface equation written with the most significant principal component scores that can be used to replace the original correlated dataset. Besides the dimensional reduction, this multiobjective programming approach has the benefit of considering the correlation among the multiple responses while generating convex Pareto frontiers to mean square error (MSE) functions. To demonstrate the procedure of the proposed approach, we used a bivariate case of AISI 52100 hardened steel turning employing wiper mixed ceramic tools. Theoretical and experimental results are convergent and confirm the effectiveness of the proposed approach.

A study of electricity price volatility for the Brazilian energy market

In the recent months, the price of the electricity in Brazil has presented a high level of volatility. As an example, the verified highest electricity price return in March 2007 was almost 260%. The volatility of a commodity plays an important role in the study of the risk management. It also improves the efficiency in parameter estimation and the accuracy in interval forecast. In this work, the Generalized Autoregressive Conditional Heteroscedastic (GARCH) model is used to study the price volatility in the Brazilian market in four geographical regions. The results have shown that the model is able to estimate the behavior of the volatility.

Multiobjective portfolio optimization of ARMA-GARCH time series based on experimental designs

The modern portfolio theory has been trying to determine how an investor might allocate assets among the possible investments options. Since the seminal contribution provided by Harry Markowitzs theory of portfolio selection, several other tools and procedures have been proposed to deal with return-risk trade-off. Furthermore, diversification across sources of returns and risks based on entropy indexes is another pivotal aspect in portfolio management. An efficient approach to model these portfolio properties with the proportion of each asset can be obtained according to mixture design of experiments. Desirability method can be applied to optimize this nonlinear multiobjective problem. Nevertheless, a tuning procedure is required, since preference articulation parameters in desirability algorithm are unknown a priori. As a result, a computer-aided desirability tuning method is proposed to find an optimal portfolio with time series of returns and risks modeled by ARMA-GARCH models. To assess the proposal feasibility, the method is tested with a heteroskedastic dataset formed by weekly world crude oil spot prices and returns. Computer-aided desirability tuning was able to enhance the global desirability by 79% in relation to the result with no tuning procedure. Display Omitted ARMA-GARCH models mean and variance of heteroskedastic time series.MDE defines asset proportions for portfolio optimization.Entropy function improves portfolio diversity.RSM sets the best weight and importance for desirability routine.Optimal portfolio is selected after computer-aided desirability tuning method.

Efficiency of tariff structure changes

This work presents the demand market share as a crucial element on the tariff establishment process. The proposed approach permits that the established tariffs are properly technical (based on the marginal cost of power supply) and efficient on the network optimization point of view. For the analysis, an econometric model of the load is considered. The results are obtained using the Energetic Company of Brasilia (CEB) data.

A hierarchical hybrid neural model with time integrators in long-term peak-load forecasting

A novel hierarchical hybrid neural model to the problem of long-term peak-load forecasting is proposed in this paper. The neural model is made up of two self-organizing map nets - one on top of the other -, and a single-layer perceptron. It has application into domains in which the context information given by former events plays a primary role. The model is compared to a multilayer perceptron. Both the hierarchical and the multilayer perceptron models are trained and assessed on load data extracted from a North-American electric utility. They are required to predict either once every week or once every month the electric peak-load during the next two years. The results are presented and evaluated in the paper.

Identifying reactive market areas

This paper discusses the problem of loss reduction by the means of reactive power redispatch. The idea is to keep the system security following a transaction. In this sense, identifying the generators most likely to play redispatch is crucial. A sensitivity-based methodology is proposed, so the generators are identified and their reactive power equations are incorporated into the ordinary power flow Jacobian, which yields a suboptimal solution. The tests are obtained with the help of the IEEE-14 bus test system with all limits taken into consideration.

An Approach to Re-dispatch Generators Based on the Modified Power Flow Jacobian Matrix

This paper focuses on the problem of re-dispatching the active power generation in order to increase the loadability of the power system. The continuation power flow is used to provide a sequence of solutions from a base case to the maximum power demand. The power flow problem solved through Newton-Raphson during the application of the continuation method is modified to include an additional equation relating the active power transmission loss and the active power generation of a critical area. This modification allows to re-dispatch the active power generation in order to reduce the transmission loss, which is shown to increase the power system loadability. The generators available to re-dispatch are identified by using sensitivity relationships, obtained as a sub-product of the continuation method. Numerical results obtained with the IEEE 57-bus test system are used to show the performance of the proposed approach.

A hybrid neural model in long-term electrical load forecasting

A novel hierarchical hybrid neural model to the problem of long-term electrical load forecasting is proposed in this paper. The neural model is made up of two self-organizing map nets — one on top of the other —, and a single-layer perceptron. It has application into domains which require time series analysis. The model is compared to a multilayer perceptron. Both the hierarchical and the multilayer perceptron models are endowed with time windows in their input layers. They are trained and assessed on load data extracted from a North-American electric utility. The models are required to predict once every week the electric peak-load and mean-load during the next two years. The results are presented and evaluated in the paper.

Forecasting financial series using clustering methods and support vector regression

This paper proposes a two-stage model for forecasting financial time series. The first stage uses clustering methods in order to segment the time series into its various contexts. The second stage makes use of support vector regressions (SVRs), one for each context, to forecast future values of the series. The series used in the experiments is composed of values of an equity fund of a Brazilian bank. The proposed model is compared to a hierarchical model (HM) presented in the literature. In this series, the HM presented prediction results superior to both a support vector machine (SVM) and a multilayer perceptron (MLP) models. The experiments show that the proposed model is superior to HM, reducing the forecasting error of the HM by 32%. This means that the proposed model is also superior to the SVM and MLP models. An analysis of the construction and use of clusters associated with a series volatility study shows that data obtained from only one type of volatility (low or high) are enough to provide sufficient knowledge to the model so that it is able to forecast future values with good accuracy. Another analysis on the quality of the clusters formed by the model shows that each cluster carries different information about the series. Furthermore, there is always a group of SVRs capable of making adequate forecasts and, for the most part, the SVR used in forecasting is a SVR belonging to this group.

Impact of Dynamic Thermal Limits of Transmission Lines on Power System Operation

This paper discusses a methodology to determine the ampacity of transmission lines based on their dynamic thermal limits. In this sense, temperature, wind speed, season conditions, and system loading are taken into account, rendering a methodology that determines, for each operating point and weather condition, the actual thermal limit of a transmission line. The proposed technique is tested in a real scenario and compared with the technique currently employed for this purpose.