<p>Naučna rasprava izložena u ovoj tezi bavi se analizom<br />kvaliteta električne energije. Visok nivo električne<br />energije podrazumeva da su napon napajanja i struja<br />potrošača idealne sinusoide sa tačno određenom<br />amplitudom i učestanošću. Bilo kakva odstupanja od<br />idealnog nazivaju se izobličenja i najčešće se<br />karakterišu sa harmonicima.<br />Poslednjih godina dolazi do naglog razvoja<br />poluprovodničkih komponenata. Takve komponente su<br />uticale na ubrzan razvoj snažnih uređaja energetske<br />elektronike. Ti uređaji su nelinearnog karaktera, što<br />dovodi do pojave harmonika u signalima napona i struja<br />elektroenergetskog sistema. Prvi problem kojim se bavi<br />ova teza je analiza talasnih oblika struja ispravljača.<br />Metode primenjene za analizu su wavelet transformacija<br />(VT) i modulated overlapped transformacija (MLT). MLT<br />nadoknađuje nedostatak VT da dekomponuje signal u<br />odgovarajuće podopsege koji mogu sadržati i više<br />harmonika i daje tačnu informaciju o svakom harmoniku.<br />Obe metode su pogodne za offline analizu. Za online analizu<br />predložen je hibridni metod baziran na diskretnoj<br />Furijeovoj transformaciji (DFT) i adaptivnom pojasnom<br />filteru (EPLL). Hibridni metod je zadržao dinamički<br />odziv DFT-a, dok je EPLL obezbedio sinhronizaciju sa<br />osnovnom učestanošću sistema. Hibridni metod daje<br />dovoljno tačnu informaciju o osnovnom i višim<br />harmonicima samo ako su njihove učestanosti ceolobrojni<br />umnožak učestanosti osnovnog harmonika. U slučaju pojave<br />interharmonika, odnosno kada taj odnos više ne važi,<br />hibridni metod ne daje tačne rezultate. Za analizu takvih<br />signala predložen je novi metod, koji je baziran na<br />adaptivnom diskretnom pojasnom filteru (ANF) t.j. metod<br />koristi diskretni pojasni filter za modelovanje<br />harmonijskih komponenata u ulaznom signalu, dok se<br />prošireni Kalmanov filter (EKF) koristi kao adaptivni<br />mehanizam. Novi metod je preuzeo osobinu ANF-a da može<br />adaptivno da prati promene učestanosti i osobinu EKF-a<br />da ima bolji dinamički odziv. Metode su implementirane<br />na digitalnom procesoru za obradu signala i upoređene sa<br />postojećim metodama. Metode pokazuju prednosti u odnosu<br />na druge metode.</p> / <p>Scientific research in this thesis discusses power quality<br />analysis. High power quality assumes that both the voltage<br />power supply and the load current are ideal sinusoidal signals<br />with a precisely defined amplitude and frequency. Any<br />deviations from this ideal vaweform are considered as distortion<br />and are characterised by harmonics.<br />Over the last few decades, there has been a rapid development<br />of semiconductor components. Such components made an<br />impact on the fast development of power electronics devices.<br />These devices are nonlinear, introducing harmonics in both<br />voltage and current of the power grid. The first issue researched<br />in this thesis is the analysis of the rectifier voltage and current<br />waveforms. Methods used for the analysis are the wavelet<br />transform (WT) and the modulated overlapped transform (MLT).<br />The MLT overcomes the drawback of the WT, which<br />decomposes the signal into subbands that can contain more<br />harmonics, and gives accurate information about every<br />harmonic. Both methods are suitable for offline analysis. For<br />online analysis, a hybrid method is proposed, based on the<br />discrete Fourier transform (DFT) and the adaptive notch filter<br />(EPLL). The hybrid method retains a good dynamic response of<br />the DFT whereas the EPLL provides a synchronisation with the<br />fundamental system frequency. The hybrid method provides<br />accurate information on the fundamental and the higher<br />harmonics only if their frequencies are integer multiples of the<br />fundamental frequency. In the case of interharmonics, i.e. when<br />this integer ratio is not valid, the hybrid method does not provide<br />accurate results. In order to analyse such signals, a new<br />method is proposed. It is based on discrete adaptive notch filter<br />(ANF), i.e. the method uses a discrete notch filter for modeling<br />the harmonic components in the input signal, whereas an<br />Extended Kalman Filter (EKF) is used as an adaptation<br />algorithm. The adaptive notch Kalman filter inherited the<br />property of the ANF to adaptively track changes in the<br />frequency and the property of the EKF to have a faster dynamic<br />response. Methods have been implemented in a digital signal<br />processor and compared with the existing ones. The methods<br />show advantages compared to other methods.</p>

This thesis is related to the field of digital signal processing; where the aim of this research is to develop features of an infinite impulse response adaptive notch filter capable of tracking multiple complex sinusoid signals. Adaptive notch filters are commonly used in: Radar, Sonar, and Communication systems, and have the ability to track the frequencies of real or complex sinusoid signals; thus removing noise from an estimate, and enhancing the performance of a system. This research programme began by implementing four currently proposed adaptive notch structures. These structures were simulated and compared: for tracking between two and four signals; however, in their current form they are only capable of tracking real sinusoid signals. Next, one of these structures is developed further, to facilitate the ability to track complex sinusoid signals. This original structure gives superior performance over Regalia's comparable structure under certain conditions, which has been proven by simulations and results. Complex adaptive notch filter structures generally contain two parameters: the first tracks a target frequency, then the second controls the adaptive notch filter's bandwidth. This thesis develops the notch filter, so that the bandwidth parameter can be adapted via a method of steepest ascent; and also investigates tracking complex-valued chirp signals. Lastly, stochastic search methods are considered; and particle swarm optimisation has been applied to reinitialise an adaptive notch filter, when tracking two signals; thus more quickly locating an unknown frequency, after the frequency of the complex sinusoid signal jumps.

Hybrid stepper motors are a common type of electric motor used throughout industry thanks to its low-cost, high torque at low speed and open loop positioning capabilities. However, a closed loop control is often required for industrial applications with high precision requirements. The closed loop control can also be used to lower the power consumption of the motor and ensure that stalls are avoided. It is quite common to utilise a large and costly position encoder or resolver to feedback the position signal to the control logic. This thesis has explored the possibility of using a low-cost position sensor based on Hall elements. Additionally, a sensorless estimation algorithm, using only stator winding measurements, has been investigated both as a competitive alternative and as a possible complement to the position sensor. The thesis work summarises and discusses previous research attempts to adequately measure or estimate and control the hybrid stepper motors position and load angle without using a typical encoder or resolver. Qualitative results have been produced through simulations prior to implementation and experimental testing. The readings from the position sensor is subject to noise, owing to its resolution and construction. The position signal has been successfully filtered, improving its accuracy from 0.56° to 0.25°. The output from the sensorless estimation algorithm is subject to non-linear errors caused by errors in phase voltage measurements and processing of velocity changes. However, the dynamics are reliable at constant speeds and could be used for position control.