Electricity price forecasting using neural networks with an improved iterative training algorithm

In a competitive electricity market, the forecasting of energy prices is an important activity for all market participants either for developing bidding strategies or for making investment decisions. In this article, a new forecasting strategy is proposed for short-term prediction of the electricity price, which is a complex quantity with nonlinear, volatile and time-dependent behaviour. Our forecast strategy includes two novelties: a new two-stage feature selection algorithm and a new iterative training algorithm. The feature selection algorithm has two filtering stages to remove irrelevant and redundant candidate inputs, respectively. This algorithm is based on mutual information and correlation analysis. The improved iterative training algorithm is composed of two neural networks in which the output of the first neural network is one of the inputs to the second. The overall proposed strategy is applied to the Pennsylvania–New Jersey–Maryland (PJM) electricity markets and compared with some of the most recent price forecast methods.     

Linear Friction Welding of Dissimilar Materials 316L Stainless Steel to Zircaloy-4

Metallurgical and Materials Transactions A - In the nuclear industry, there are a number of applications where the transition of stainless steel to Zircaloy is of technological importance. However,...     

Substrate Integrated Waveguide Corrugated Horn Antenna

Wireless Personal Communications - In this paper, a substrate integrated waveguide (SIW) H-plane horn antenna is proposed. We first optimize the antenna in size to obtain the maximum gain. The SIW...     

Surface Runoff from Full-Scale Coal Combustion Product Pavements during Accelerated Loading

In this study, the release of metals and metalloids from full-scale portland cement concrete pavements containing coal combustion products (CCPs) was evaluated by laboratory leaching tests and accelerated loading of full-scale pavement sections under well-controlled conditions. An equivalent of 20 years of highway traffic loading was simulated at the OSU/OU Accelerated Pavement Load Facility (APLF). Three types of portland cement concrete driving surface layers were tested, including a control section [i.e., ordinary portland cement (PC) concrete] containing no fly ash and two sections in which fly ash was substituted for a fraction of the cement; i.e., 30% fly ash (FA30) and 50% fly ash (FA50). In general, the concentrations of minor and trace elements were higher in the toxicity characteristic leaching procedure (TCLP) leachates than in the leachates obtained from synthetic precipitation leaching procedure and ASTM leaching procedures. Importantly, none of the leachate concentrations exceeded the TCLP limits or primary drinking water standards. Surface runoff monitoring results showed the highest release rates of inorganic elements from the FA50 concrete pavement, whereas there were little differences in release rates between PC and FA30 concretes. The release of elements generally decreased with increasing pavement loading. Except for Cr, elements were released as particulates (>0.45?μm) rather than dissolved constituents. The incorporation of fly ash in the PC cement concrete pavements examined in this study resulted in little or no deleterious environmental impact from the leaching of inorganic elements over the lifetime of the pavement system.     

Motion detection in CT images with a novel fast technique

Different factors may influence the image quality of the first generation computed tomography (CT) system (single-source—single-detector) and one of these factors is the object motion. For studying this effect on image quality, an industrial CT system was designed and developed. Several experiments were performed with different axial and rotational motion of a phantom. The quality of reconstructed images was compared by computing the RMSE of each image. Also, in this paper a processing technique is presented that has the potential to detect the object motion during the acquisition process and to correct it by rescanning from motion point. The presented results can be extended to other medical and industrial applications.     

Predicting Chaotic Time Series Using Neural and Neurofuzzy Models: A Comparative Study

The prediction accuracy and generalization ability of neural/neurofuzzy models for chaotic time series prediction highly depends on employed network model as well as learning algorithm. In this study, several neural and neurofuzzy models with different learning algorithms are examined for prediction of several benchmark chaotic systems and time series. The prediction performance of locally linear neurofuzzy models with recently developed Locally Linear Model Tree (LoLiMoT) learning algorithm is compared with that of Radial Basis Function (RBF) neural network with Orthogonal Least Squares (OLS) learning algorithm, MultiLayer Perceptron neural network with error back-propagation learning algorithm, and Adaptive Network based Fuzzy Inference System. Particularly, cross validation techniques based on the evaluation of error indices on multiple validation sets is utilized to optimize the number of neurons and to prevent over fitting in the incremental learning algorithms. To make a fair comparison between neural and neurofuzzy models, they are compared at their best structure based on their prediction accuracy, generalization, and computational complexity. The experiments are basically designed to analyze the generalization capability and accuracy of the learning techniques when dealing with limited number of training samples from deterministic chaotic time series, but the effect of noise on the performance of the techniques is also considered. Various chaotic systems and time series including Lorenz system, Mackey-Glass chaotic equation, Henon map, AE geomagnetic activity index, and sunspot numbers are examined as case studies. The obtained results indicate the superior performance of incremental learning algorithms and their respective networks, such as, OLS for RBF network and LoLiMoT for locally linear neurofuzzy model.