An analysis of low-level feature space for emotion recognition from the speech is presented. The main goal was to determine how the statistical properties computed from contours of low-level features influence the emotion recognition from speech signals. We have conducted several experiments to reduce and tune our initial feature set and to configure the classification stage. In the process of analysis of the audio feature space, we have employed the univariate feature selection using the chi-squared test. Then, in the first stage of classification, a default set of parameters was selected for every classifier. For the classifier that obtained the best results with the default settings, the hyperparameter tuning using cross-validation was exploited. In the result, we compared the classification results for two different languages to find out the difference between emotional states expressed in spoken sentences. The results show that from an initial feature set containing 3198 attributes we have obtained the dimensionality reduction about 80% using feature selection algorithm. The most dominant attributes selected at this stage based on the mel and bark frequency scales filterbanks with its variability described mainly by variance, median absolute deviation and standard and average deviations. Finally, the classification accuracy using tuned SVM classifier was equal to 72.5% and 88.27% for emotional spoken sentences in Polish and German languages, respectively.

The design and performance analysis of a 1310/1550-nm wavelength division demultiplexer with tapered geometry based on InP/InGaAsP multimode interference (MMI) coupler has been carried out. Wavelength response of demultiplexer of conventional MMI and tapered input and tapered output (tapered I/O) waveguides geometry of the MMI have been discussed. The demultiplexing function has been first performed by choosing a suitable refractive index of the guiding region and geometrical parameters such as the width and length of MMI structure have been achieved. Access width of tapered I/O waveguides have been adjusted to give a low insertion loss (IL) and high extinction ratio (ER) for the considered wavelengths of 1310 nm and 1550 nm. The total size of the demultiplexer has been significantly reduced over the existing MMI devices. Numerical simulations with finite difference beam propagation method are applied to design and optimize the operation of the proposed demultiplexer.

Heat transfer in steady free convection from differentially heated cylinders enclosed in a rectangular duct filled with Bingham plastic fluids has been solved numerically for the ranges of the dimensionless groups as, Rayleigh number, 10 ² ≤ Ra ≤ 10 ⁶; Prandtl number, 10 ≤ Pr ≤ 100 and, Bingham number, 0 ≤ Bn ≤ 50 for aspect ratios AR = 0.5, 0.6, 0.7, 0.8, 0.9 and 2. The streamlines, isotherm contours, yield surfaces, local and average Nusselt numbers were analysed and discussed. It is found that as the aspect ratio of the enclosure increases from 0.5 to 0.9, the average Nusselt number on the surface of the hot cylinder increases as a larger amount of fluid takes part in convection. Moreover, at sufficiently large Bingham numbers, yield stress forces dominate over buoyancy causing the flow to cease and thus the Nusselt number approaches its conduction limit. Finally, the Nusselt number approaches its conduction limit once the maximum Bingham number is reached.

This paper presents a predictive torque and flux control algorithm for the synchronous reluctance machine. The algorithm performs a voltage space phasor pre-selection, followed by the computation of the switching instants for the optimum switching space phasors, with the advantages of inherently constant switching frequency and time equidistant implementation on a DSP based system. The criteria used to choose the appropriate voltage space phasor depend on the state of the machine and the deviations of torque and flux at the end of the cycle. The model of the machine has been developed on a d-q frame of coordinates attached to the rotor and takes into account the magnetic saturation in both d-q axes and the cross saturation phenomenon between both axes. Therefore, a very good approximation of this effect is achieved and the performance of the machine is improved. Several simulations and experimental results using a DSP and a commercially available machine show the validity of the proposed control scheme.