The paper aims at comparing forecast ability of VAR/VEC models with a non-changing covariance matrix and two classes of Bayesian Vector Error Correction – Stochastic Volatility (VEC-SV) models, which combine the VEC representation of a VAR structure with stochastic volatility, represented by the Multiplicative Stochastic Factor (MSF) process, the SBEKK form or the MSF-SBEKK specification.
Based on macro-data coming from the Polish economy (time series of unemployment, inflation and interest rates) we evaluate predictive density functions employing of such measures as log predictive density score, continuous rank probability score, energy score, probability integral transform. Each of them takes account of different feature of the obtained predictive density functions.
A speaker recognition system based on joint factor analysis (JFA) is proposed to improve whispering speakers’ recognition rate under channel mismatch. The system estimated separately the eigenvoice and the eigenchannel before calculating the corresponding speaker and the channel factors. Finally, a channel-free speaker model was built to describe accurately a speaker using model compensation. The test results from the whispered speech databases obtained under eight different channels showed that the correct recognition rate of a recognition system based on JFA was higher than that of the Gaussian Mixture Model-Universal Background Model. In particular, the recognition rate in cellphone channel tests increased significantly.
Dry electrostatic precipitators (ESPs) are widely used for purification of exhaust gases in industrial applications. Maintenance of their high efficiency depends primarily on periodical cleaning of the collecting electrodes (CEs). Dust removal (regeneration of CEs) is realized by inducing periodical vibrations of the electrodes. The paper presents results of vibration modelling of a system of CEs; the results were obtained by means of the finite element method, the hybrid finite element method, the finite strip method and a model formulated using Abaqus. Numerical results are compared with those obtained from experimental measurements. Conclusions concerning numerical effectiveness and exactness of the methods are formulated and reasons for differences are discussed.
Buildings consume half of all energy use and are also responsible for a similar proportion of carbon dioxide emission. The heat transfer across the building envelope - the shell of a house that separates the inside and outside - should generally be minimized. In the paper validation and verification based on Building Energy Simulation Test (BESTEST) of Energy3D computer code is presented. Next, computations performed by means of Energy 3D and Energy Plus for BESTEST building are compared. In the last part of the paper results for computations for real building are presented. Program Energy 3D proved to be an excellent tool for qualitative and quantitative analysis of buildings with respect to energy consumption.
According to metrological guidelines and specific legal requirements, every smart electronic electricity meter has to be constantly verified after pre-defined regular time intervals. The problem is that in most cases these pre-defined time intervals are based on some previous experience or empirical knowledge and rarely on scientifically sound data. Since the verification itself is a costly procedure it would be advantageous to put more effort into defining the required verification periods. Therefore, a fixed verification interval, recommended by various internal documents, standardised evaluation procedures and national legislation, could be technically and scientifically more justified and consequently more appropriate and trustworthy for the end user. This paper describes an experiment to determine the effect of alternating temperature and humidity and constant high current on a smart electronic electricity meter’s measurement accuracy. Based on an analysis of these effects it is proposed that the current fixed verification interval could be revised, taking into account also different climatic influence. The findings of this work could influence a new standardized procedure in respect of a meter’s verification interval.
Numerical simulation is an economical and effective method in the field of marine engineering. The dynamics of mooring cables has been analysed by a numerical simulation code that was created on a basis of a new element frame. This paper aims at verifying the accuracy of the numerical simulation code through comparisons with both the real experiments and a commercial simulation code. The real experiments are carried out with a catenary chain mooring in a water tank. The experimental results match the simulation results by the numerical simulation code well. Additionally, a virtual simulation of a large size chain mooring in ocean is carried out by both the numerical simulation code and a commercial simulation code. The simulation results by the numerical simulation code match those by the commercial simulation code well. Thus, the accuracy of the numerical simulation code for underwater chain mooring is verified by both the real experiments and commercial simulation code.
The rigid finite element method (RFEM) has been used mainly for modelling systems with beam-like links. This paper deals with modelling of a single set of electrodes consisting of an upper beam with electrodes, which are shells with complicated shapes, and an anvil beam. Discretisation of the whole system, both the beams and the electrodes, is carried out by means of the rigid finite element method. The results of calculations concerned with free vibrations of the plates are compared with those obtained from a commercial package of the finite element method (FEM), while forced vibrations of the set of electrodes are compared with those obtained by means of the hybrid finite element method (HFEM) and experimental measurements obtained on a special test stand.
In the paper, the author analyses a model of a ring pack motion on an oil film. The local thickness of the oil film can be compared to the height of the combined surface roughness of a cylinder liner and piston rings. Equations describing the mixed lubrication problem based on the empirical mathematical model formulated in works by Patir, Cheng [6, 7] and Greenwood, Tripp [3] have been combined [13] and used in this paper. A model of a gas flow through the labyrinth seal of piston rings has been developed [14,16]. In addition, models of ring twist effects and axial ring motion in piston grooves have been applied [15,16].
In contrast to the previous papers of the author, an experimental verification of the main parts of developed mathematical model and software has been presented. A relatively good compatibility between the experimental measurements and calculated results has been achieved.
In this work, the author presents experimental verification of numerical simulation of projectile impact on constructional shields. The experimental tests were performed at a unified test stand to investigate ballistic resistance of materials in field conditions. The stand was developed at the Polish Naval Academy in Gdynia, and then patented. The design of this test stand was based on construction of a ballistic pendulum, fitted to measure: impact force, turn angle of the ballistic pendulum χ, impact velocity and residual velocity of the projectile. All the measurement data were transmitted to a digital oscilloscope and a personal computer. The ballistic velocity of
the shield of VBL[R] – defined according to Recht’s and Ipson’s method, was compared with VBL[Z] and VBL[Z1] – determined according to the author’s method. Verification of numerically simulated ballistic velocity VRO versus the before-mentioned velocity
was carried out at the 10GHMBA-E620T steel shields impacted by 12.7 mm type B-32 projectiles. The introduced method can be used for determining ballistic thickness hBL and ballistic velocity VBL for both homogeneous plates as well as multi-layered constructional shields.
Biometrics provide an alternative to passwords and pins for authentication. The emergence of machine learning algorithms provides an easy and economical solution to authentication problems. The phases of speaker verification protocol are training, enrollment of speakers and evaluation of unknown voice. In this paper, we addressed text independent speaker verification using Siamese convolutional network. Siamese networks are twin networks with shared weights. Feature space can be learnt easily by training these networks even if similar observations are placed in proximity. Extracted features from Siamese then can be classified using difference or correlation measures. We have implemented a customized scoring scheme that utilizes Siamese’ capability of applying distance measures with the convolutional learning. Experiments made on cross language audios of multi-lingual speakers confirm the capability of our architecture to handle gender, age and language independent speaker verification. Moreover, our designed Siamese network, SpeakerNet, provided better results than the existing speaker verification approaches by decreasing the equal error rate to 0.02.