George E. Nasr

Univariate modeling and forecasting of energy consumption : the case of electricity in Lebanon

In Lebanon, electric power is becoming the main energy form relied upon in all economic sectors of the country. Also, the time series of electrical energy consumption in Lebanon is unique due to intermittent power outages and increasing demand. Given these facts, it is critical to model and forecast electrical energy consumption. The aim of this study is to investigate different univariate-modeling methodologies and try, at least, a one-step ahead forecast for monthly electric energy consumption in Lebanon. Three univariate models are used, namely, the autoregressive, the autoregressive integrated moving average (ARIMA) and a novel configuration combining an AR(1) with a highpass filter. The forecasting performance of each model is assessed using different measures. The AR(1)/highpass filter model yields the best forecast for this peculiar energy data.

Econometric modeling of electricity consumption in post-war Lebanon

This paper applies econometric models to investigate determinants of electrical energy consumption in post-war Lebanon. The impact of the Gross Domestic Product (GDP), proxied by total imports (TI), and degree days (DD) on electricity consumption is investigated over different time spans covering the period from 1993 to 1997. The time spans are chosen according to the rationing level of electricity supply. For the 1993–1994 time span, TI is found to be a significant determinant of energy consumption, whereas, DD has a negative correlation. This inconsistency might be attributed to an extensive rationing policy followed during this period. For the 1995–1997 time span which includes reduced rationing period (1995), all electrical energy consumption determinants are found to be significant at the 5% significance level. Analysis results for the rationing free 1996–1997 time span also show the significance of TI and DD at the 5% level. Furthermore, cointegration analysis for the 1995–1997 and 1996–1997 subsets reveals the existence of a long-run relationship between all variables. In addition, error correction models for both subsets are developed to predict short-run dynamics. Finally, statistical performance measures such as mean square error, mean average deviation and mean average percentage error are presented for all models.

Backpropagation neural networks for modeling gasoline consumption

This paper presents an artificial neural network (ANN) approach to gasoline consumption (GC) forecasting in Lebanon. In order to provide the forecasted gasoline consumption, the ANN interpolates among the GC and its determinants in a training data set. In this study, four ANN models are presented and implemented on real GC data. The first model is a univariate model based on past consumption values. The second model is a multivariate model based on GC time series and price (P). The third model is also a multivariate model based on GC and car registration (CR). Finally, the fourth model combines GC, P and CR. Forecasting performance measures, such as mean square errors and mean absolute deviations, are presented for all models.

Compensation of axle-generator errors due to wheel slip and slide

Significant errors of train axle generators (tachometers) are due to wheel slip and slide. An algorithm is designed to compensate for these errors. The algorithm identifies the wheel slip and slide by examining the variation of the processed vehicle longitudinal acceleration. Whenever wheel slip/slide is identified, then the vehicle speed is adjusted if a certain condition is met. The adjustment is a simple linear interpolation between the two speed values recorded before and after wheel slip/slide detection. In addition, a speed and acceleration observer using a Kalman filter is implemented. Experimental results using three different axle encoders aboard a freight train are provided to illustrate the performance of the proposed algorithm.

Sensitivity of discrete‐time Kalman filter to statistical modeling errors

The optimum filtering results of Kalman filtering for linear dynamic systems require an exact knowledge of the process noise covariance matrix Qk, the measurement noise covariance matrix Rk and the initial error covariance matrix P0. In a number of practical solutions, Qk, Rk and P0, are either unknown or are known only approximately. In this paper the sensitivity due to a class of errors in statistical modelling employing a Kalman Filter is discussed. In particular, we present a special case where it is shown that Kalman filter gains can be insensitive to scaling of covariance matrices. Some basic results are derived to describe the mutual relations among the three covariance matrices (actual and perturbed covariance matrices), their respective Kalman gain Kk and the error covariance matrices Pk. It is also shown that system modelling errors, particularly scaling errors of the input matrix, do not perturb the Kalman gain. A numerical example is presented to illustrate the theoretical results, and also to show the Kalman gain insensitivity to less restrictive statistical uncertainties in an approximate sense. Copyright

Econometric Modeling of Gasoline Consumption: A Cointegration Analysis

Abstract This article applies econometric models to investigate determinants of gasoline consumption (GC) in postwar Lebanon (1993–1999). The impact of gasoline price (P) and car registration (CR) on gasoline consumption is investigated through three models, namely, the static, autoregressive, and partial adjustment models. Analysis results showed the statistical significance of the price, at the 10% level, in affecting gasoline consumption and the insignificance of the car registration time series. Furthermore, given that regression models may produce spurious results, if time series are non-stationary, the GC, P, and CR were tested for order of integration using the Dickey-Fuller (DF) and the augmented Dickey-Fuller (ADF) tests. Cointegration analysis, using the Johansen and the Engle and Yoo test methods, revealed the existence of a long-run relationship between all variables. Moreover, an error correction model is developed to predict short-run dynamics. Finally, statistical performance measures, such as mean square error, mean average deviation, and mean average percentage error, are presented for all models.

A GUI-based artificial neural network simulator

In this paper, a Java-based artificial neural network simulator is presented. Requirements for an educational model are also introduced and an implementation procedure is then depicted. Feature comparison between this model and some well known similar tools such as Xerion and SNNS are also performed. Finally, extension possibilities, including a client-server approach suitable for education and classroom approach are also discussed.