Dragan Kukolj

Identification of complex systems based on neural and Takagi-Sugeno fuzzy model

The paper describes a neuro-fuzzy identification approach, which uses numerical data as a starting point. The proposed method generates a Takagi-Sugeno fuzzy model, characterized with transparency, high accuracy and a small number of rules. The process of self-organizing of the identification model consists of three phases: clustering of the input-output space using a self-organized neural network; determination of the parameters of the consequent part of a rule from over-determined batch least-squares formulation of the problem, using singular value decomposition algorithm; and on-line adaptation of these parameters using recursive least-squares method. The verification of the proposed identification approach is provided using four different problems: two benchmark identification problems, speed estimation for a DC motor drive, and estimation of the temperature in a tunnel furnace for clay baking.

Design of adaptive Takagi–Sugeno–Kang fuzzy models

Abstract The paper describes a method of fuzzy model generation using numerical data as a starting point. The algorithm generates a Takagi–Sugeno–Kang fuzzy model, characterised with transparency, high accuracy and small number of rules. The training algorithm consists of three steps: partitioning of the input–output space using a fuzzy clustering method; determination of parameters of the consequent part of a rule from over-determined batch least-squares (LS) formulation of the problem, using singular value decomposition algorithm; and adaptation of these parameters using recursive least-squares method. Three illustrative well-known benchmark modelling problems serve the purpose of demonstrating the performance of the generated models. The achievable performance is compared with similar existing models, available in literature.

Salient Motion Features for Video Quality Assessment

Design of algorithms that are able to estimate video quality as perceived by human observers is of interest for a number of applications. Depending on the video content, the artifacts introduced by the coding process can be more or less pronounced and diversely affect the quality of videos, as estimated by humans. While it is well understood that motion affects both human attention and coding quality, this relationship has only recently started gaining attention among the research community, when video quality assessment (VQA) is concerned. In this paper, the effect of calculating several objective measure features, related to video coding artifacts, separately for salient motion and other regions of the frames of the sequence is examined. In addition, we propose a new scheme for quality assessment of coded video streams, which takes into account salient motion. Standardized procedure has been used to calculate the Mean Opinion Score (MOS), based on experiments conducted with a group of non-expert observers viewing standard definition (SD) sequences. MOS measurements were taken for nine different SD sequences, coded using MPEG-2 at five different bit-rates. Eighteen different published approaches related to measuring the amount of coding artifacts objectively on a single-frame basis were implemented. Additional features describing the intensity of salient motion in the frames, as well as the intensity of coding artifacts in the salient motion regions were proposed. Automatic feature selection was performed to determine the subset of features most correlated to video quality. The results show that salient-motion-related features enhance prediction and indicate that the presence of blocking effect artifacts and blurring in the salient regions and variance and intensity of temporal changes in non-salient regions influence the perceived video quality.

Design of a PID-like compound fuzzy logic controller

Abstract The paper describes a novel method for the design of a fuzzy logic controller (FLC) with near-optimal performance for a variety of operating conditions. The approach is based on the analysis of the system behaviour in the error state-space. The final control structure, in a form of a compound FLC, is arrived at in two stages. The first stage encompasses design and tuning of a PID-like fuzzy controller. The second stage consists of placing an additional fuzzy controller, of a structure similar to that of the first one, in parallel with the PID-like fuzzy controller designed in the first stage. The resulting compound controller is characterised with high performance in the wide range of operating conditions, and with small number of parameters that can be adjusted using simple optimisation methods. The controller is developed and tested for a plant comprising a vector controlled induction motor drive.

Option Pricing With Modular Neural Networks

This paper investigates a nonparametric modular neural network (MNN) model to price the S&P-500 European call options. The modules are based on time to maturity and moneyness of the options. The option price function of interest is homogeneous of degree one with respect to the underlying index price and the strike price. When compared to an array of parametric and nonparametric models, the MNN method consistently exerts superior out-of-sample pricing performance. We conclude that modularity improves the generalization properties of standard feedforward neural network option pricing models (with and without the homogeneity hint).

Indoor fingerprint localization in WSN environment based on neural network

The indoor localization is an actual problem because there are more and more application areas. New technical solutions are available, which have contributed to the indoor localization researches. In this work Fingerprint (FP) localization methodology applied in the experimental indoor environment is presented. The Wireless Sensor Network technology (WSN) is used in real environment, which provided the necessary measurement results to the FP localization. For the processing Received Signal Strength Indicator (RSSI) and for determining the position the neural network model is used. The RSSI values used for the learning of the neural network are preprocessed (mean, median, standard deviation) in order to increase the accuracy of the system. The type of the neural network is a feed-forward network. During obtain learning different algorithms were applied. The mean square error of Euclidean distance between calculated and real coordinates and the histogram of precision were used to determine the accuracy of the neural network.

Design of the speed controller for sensorless electric drives based on AI techniques: a comparative study

Abstract The paper investigates applicability of different artificial intelligence (AI) techniques in the design of a speed controller for electric drives. A speed-sensorless drive system is considered. A controller structure consisting of a load torque observer, a speed estimator and a speed predictor is developed. Next, different AI based approaches to speed controller design are investigated. The speed controllers based on (1) feed-forward neural network, (2) neuro-fuzzy network, and (3) self-organising Takagi–Sugeno (TS) rule based model are designed. A comparative analysis of the drive behaviour with these three types of AI based speed controllers is performed. In addition, a comparison is made with respect to the drive performance obtained with a conventional optimised PI controller. A detailed simulation study of a number of transients indicates that the best performance, in terms of accuracy and computational complexity, is offered by the self-organising Takagi–Sugeno controller. The controllers are developed and tested for a plant comprising a variable-speed separately excited DC motor.

A human detection method f or residential smart energy systems based on Zigbee RSSI changes

In this article, the device-free human presence detection method based on radio signal strength variations is proposed. The method exploits the known fact that human body interferes with radio signals by causing fading and shadowing effects. Introduced irregularities in the radio propagation pattern indicate possible presence of a human. The proposed method is incorporated into the existing platform for intelligent residential energy management. As opposed to conventional solutions which utilize a complex set of sensors for human detection, the proposed approach achieves the same only by analyzing and quantifying radio signal strength variations incorporated in messages exchanged between 2.4 GHz radio transceivers. One of the key benefits of the proposed solution is the integration of the detection algorithm into the smart power outlets and smart light switches. Such an approach improves interactions in smart home systems, enables intelligent power consumption management and low installation cost.

DIBR synthesized image quality assessment based on morphological wavelets

Most of the Depth Image Based Rendering (DIBR) techniques produce synthesized images which contain nonuniform geometric distortions affecting edges coherency. This type of distortions are challenging for common image quality metrics. Morphological filters maintain important geometric information such as edges across different resolution levels. In this paper, morphological wavelet peak signal-to-noise ratio measure, MW-PSNR, based on morphological wavelet decomposition is proposed to tackle the evaluation of DIBR synthesized images. It is shown that MW-PSNR achieves much higher correlation with human judgment compared to the state-of-the-art image quality measures in this context.

System Design for Passive Human Detection Using Principal Components of the Signal Strength Space

Radio irregularity phenomenon is often considered as a shortcoming of wireless sensor networks. In this paper, the radio irregularity is regarded as a benefit of wireless networks. The proposed novel method for human presence detection, applied to the monitoring system for residential energy management, utilizes radio irregularity for an efficient human presence detection which increases user awareness and automates the power control. The method monitors 2.4GHz wireless (Zigbee) smart outlets and the communication signals between them to detect changes of the received signal strength indicator (RSSI) and its variations compared to the expected mean values. The signal strength variation is significantly increased due to human presence within the signal propagation path. By applying the algorithm for principal components analysis to the set of RSSI samples obtained from radio communication links within the detection area, it is possible to recognize a human presence. There is no need for additional presence sensors installation, because the proposed method applied to the smart outlets network is quite satisfactory to extend the entire system with the detection capability. Compared to the conventional sensor networks, the presented solution preserves the pervasiveness of smart energy and smart home systems, high level of sensorial intelligence, simplicity and low installation costs.