A Genetic Algorithm Approach for Three-Phase Harmonic Mitigation Filter Design

  • Hazem Zubi

Student thesis: Doctoral ThesisPhD


In industry, adjustable speed drives (ASDs) are widely employed in driving AC motors for variable speed applications due to the high performance and high energy efficiency obtained in such systems. However, ASDs have an impact on the power quality and utilisation of AC power feeds by injecting current harmonics and causing resonances, additional losses, and voltage distortion at the point of common coupling. Due to these problems, electric power utilities have established stringent rules and regulations to limit the effects of this distortion. As a result, efficient, reliable, and economical harmonic mitigation techniques must now be implemented in practical systems to achieve compliance at reasonable cost.A variety of techniques exist to control the harmonic current injected by ASDs, and allow three-phase AC-line-connected medium-power systems to meet stringent power quality standards. Of these, the broadband harmonic passive filter deserves special attention because of its good harmonic mitigation and reactive power compensation abilities, and low cost. It is also relatively free from harmonic resonance problems, has relatively simple structural complexity and involves considerably less engineering effort when compared to systems of single tuned shunt passive filters or active filters and active rectifier solutions.In this thesis, passive broadband harmonic filters are investigated. In particular, the improved broadband filter (IBF) which has superior overall performance and examples of its application are increasing rapidly. During this research project, the IBF operating principle is reviewed and its design principles are established. As the main disadvantage of most passive harmonic filters is the large-sized components, the first proposed design attempts to optimize the size of the filter components (L and C) utilized in the existing IBF topology. The second proposed design attempts to optimize the number and then the size of filter components resulting in an Advanced Broadband passive Filter (ABF) novel structure.The proposed design methods are based on frequency domain modelling of the system and then using a genetic algorithm optimization technique to search foroptimal filter component values. The results obtained are compared with the results of a linear searching approach.The measured performance of the optimal filter designs (IBF and ABF) is evaluated under different loading conditions with typical levels of background voltage distortion. This involves assessing input current total harmonic distortion, input power factor, rectifier voltage regulation, efficiency, size and cost. The potential resonance problem is addressed and the influence of voltage imbalance on performance is investigated. The assessment is based on analysis, computer simulations and experimental results. The measured performance is compared to various typical passive harmonic filters for three-phase diode rectifier front-end type adjustable speed drives.Finally, the broadband filter design’s effectiveness and performance are evaluated by involving them in a standard IEEE distribution network operating under different penetration levels of connected nonlinear total loads (ASD system). The study is conducted via detailed modelling of the distribution network and the linked nonlinear loads using computer simulations.
Date of Award31 Dec 2013
Original languageEnglish
Awarding Institution
  • University of Bath
SupervisorRoderick Dunn (Supervisor) & Raj Aggarwal (Supervisor)


  • broadband filter design
  • drive
  • passive filter
  • power factor
  • rectifier
  • thd

Cite this