Artificial Intelligence based Hybrid Filters for Power Quality Improvement

Abstract

newlineThe broad utilization of power electronics loads in various types of commercial, newlineindustrial and household electrical tools causes the degradation of the quality of the power newlinesupply by distorting the supply voltage. The harmonic problems are mainly evolved newlinebecause of extensive use of nonlinear loads in industry and domestic sectors. There are newlinesome adverse effects of harmonics such as: malfunctioning of sensitive equipment, newlineresonance issues, conductors heating, power losses and reduced efficiency in distribution newlinesystem. Due to the stricter requirements as per IEEE-519 standard, the compensation of newlineharmonics is required in order to improve the power quality. Hence, to mitigate newlineharmonics, passive filters are conventionally used. But, with the increase of harmonic newlinecomponents, the design of passive filters becomes more complex and bulky. With the newlineadvancement in power electronics, the active power filter has been designed in order to newlineeliminate the demerits of passive filters. Particularly for three phase configurations, the newlinevoltage source structures give excellent results in improving the power quality. But, the newlinerating of the active filter is very high for some applications (up to 80% of load rating). newlineHence, Hybrid Shunt Active Power Filter (HSAPF) is designed using low rated Shunt newlineActive Power Filter (SAPF) and low cost Shunt Passive Filter (SPF) to enhance the newlinecompensation ability. HSAPF gives a cost effective solution by exploiting the merits and newlineeliminating the demerits of shunt passive and shunt active filters. newline Hence in this thesis, HSAPF system has been implemented to eliminate the newlineharmonics, to reduce the reactive power burden and to achieve unity power factor at the newlinesource side. The designed filter has the objective to eliminate the load generated newlineharmonics under steady state, transient, balanced, unbalanced, diode and thyristorised newlinenonlinear loading conditions using MATLAB/Simulink and real-time environments. In newlinethis work, the compensation strategies for HSAPF is developed based on source current newlineextraction approach which is different from conventional methods such as (p-q) theory or newline(id-iq) theory. Generally, the compensation process can be divided into two parts, out of newlinewhich first one is extraction of reference current which is done by DC link capacitor newlinevoltage regulation. The second part is the generation of switching pulse by PWM newlinetechnique. To extract the reference current, firstly the conventional Proportional Integral newlineController (PIC), Self-Tuned Proportional Integral Controller (STPIC) or the Fuzzy Logic newlineController (FLC) is used and the switching pulse generation is done by fixed hysteresis, newlineadaptive hysteresis, sliding mode control (SMC) and adaptive fuzzy hysteresis control. newlinevi newline Subsequently, to enhance the performance of HSAPF, Fractional Order Proportional newlineIntegral Derivative Controller (FOPIDC) is used to extract the reference current. Due to newlinethe research interest in fractional calculus, FOPIDC is tested to be more effective in newlinedifferent applications as compared to Conventional Proportional Integral Derivative newlineController (CPIDC). As we know that the Ziegler Nichols tuning method is popularly used newlinefor tuning of CPIDC and needs wide range of experimentation that generally yields newlineinadequate results. Hence, there is always possibility of better tuning of CPIDC or newlineFOPIDC to improve the controller performances. Hence, the controller gains are newlineoptimized by various optimization techniques such as: Particle Swarm Optimization newline(PSO), Grey Wolf Optimization (GWO), Hybrid PSO-GWO Optimization, Teaching newlineLearning Based Optimization (TLBO), Football Game Optimization (FGO) and Improved newlineFootball Game Optimization (IFGO). newline Subsequently, to enhance the performance of HSAPF, Fuzzy PID Controller newline(FPIDC) and Self Adaptive Fuzzy PID Controller (SAFPIDC) is proposed for reference newlinecurrent generation. Again, to assess the accuracy of the reference current estimation newlinetechnique, Kalman Filter (KF), Extended Complex Kalman Filter (ECKF) and Robust newlineExtended Complex Kalman Filter (RECKF) are used to estimate the per unit fundamental newlinecomponent of the reference current. Finally, the HSAPF uses Dual Tree-Complex newlineWavelet Transform (DT-CWT) for generating the reference currents and PIC, Type-1 and newlineType-2 Fuzzy Logic Controllers (T1FLC and T2FLC) are employed to optimize the newlineparameters of the HSAPF to improve the harmonics compensation ability and the power newlinefactor. In order to justify the better performance of the proposed controllers, simulations as newlinewell as real time case studies are presented using the above mentioned adaptive techniques newlineunder different operating scenarios