Artificial intelligence-based approaches to control the neutral current compensator and implementation of power quality conditioner for power quality enhancement

Improving Power Quality through Artificial Intelligence-based Control Techniques

Abstract

Power electronic components and electronic converters gave birth to numerous newapplications, offering unmatched comfort, flexibility and efficiency to the customers. Owing to theimprovement in technologies, large number of harmonics are introduced and they degrade theperformance of the three-phase four-wire distribution system. These effects of distortion also affect thepoint of common coupling where other loads are connected to the system. A lot of measures andmitigating techniques have been presented in the literature but each has its own drawbacks like circuitcomplexity, lack of reducing lower order harmonics, occurrence of resonance problems with lineimpedance, etc. To overcome these defects, the thesis proposes two techniques to improve the powerquality of the system. The first method aims at designing a Neutral Current Compensator and the secondmethod involves designing of Unified Power Quality Conditioner. The neutral conductor of three-phasefour-wire distribution systems have been widely employed to deliver electric power to single-phase orthree-phase loads. The neutral conductor usually carries the zero-sequence current due to theunbalanced loading among phase conductors. As more and more electronic equipment such ascomputers, copy machines and adjustable speed drives are being used, the harmonic currents drawn bytheir rectifier front ends also become significant. Especially, the zero-sequence triple (3rd, 9th, etc)harmonics accumulate in the neutral conductor, resulting in overloading of the neutral conductor andthe distribution transformer. Survey results in United States have indicated that 22.6 of the sites haveneutral current exceeding the phase currents. In order to alleviate the effects of neutral current, thethesis proposes a new technique based on neutral current compensator. The neutral currentcompensator is a series active filter which is connected in series with the neutral conductor. The mainobjective of this compensator is to eliminate the harmonic component present in the neutral conductorthere by overloading and overheating in the neutral conductor can be eliminated. By eliminating theharmonic distortion in the neutral, the phase current distortion is also reduced to a considerable extent.Rotating unit vector-based hysteresis controller is used to extract the harmonic component in theneutral conductor. The proposed method implements back propagated artificial neural network, fuzzylogic controller techniques and fuzzy solar PV based hysteresis controllers. The MATLAB simulationsoftware is used to model and simulate the circuits. The hardware model is also implemented for thecompensation of the neutral current harmonics to validate the simulation results. The effects ofnonlinear loads also have a high impact on the load voltage distortion and load current distortion.Consequently, the current drawn by the loads has a high harmonic distortion level. If this levelcontinues, then it affects the performance of the system by causing over heating of the conductors anddegrades the efficiency. The thesis prescribes a model of Unified Power Quality Conditioner to overcomethese potential problems. Synchronous Reference Frame controller is used to extract the referencesignals for the series and shunt inverter of unified power quality conditioner. The conventionalProportional Integral (PI) controllers, genetic algorithm-based tuning of PI controllers and particle swarmoptimization-based tuning of PI controllers are implemented for maintaining constant voltage across theinput of inverter circuit. Hysteresis current and voltage controllers are used to generate the switchingsignals for the series and shunt inverters. Fuzzy logic controller is implemented to control the upper andlower band of the hysteresis controller. Among the various techniques stipulated, the PSO tuned PIcontroller with fuzzy hysteresis controller provides good dynamic response by reducing the harmonicdistortion in the load current and load voltage, thereby balancing the unbalanced loads.