GENETIC ALGORITHM BASED HARMONIC EVALUATION AND FILTER PLACEMENT FOR A RAPID TRANSIT SYSTEM

Abstract

Mass rapid transit (MRT) operation can cause harmonic pollution problems in power systems. A great deal of measures towards the reduction of harmonic components should be investigated. Before implementing a harmonic compensation scheme, levels of harmonic distortion under all foreseeable operating modes of the study system should be evaluated. Several harmonics criteria can be adopted to assess the levels and harmonic distribution around the system. The extent of harmonic interference with other devices should be studied. Rectifiers and inverters, which bridge between the power supply and MRT systems, produce most of the harmonics in the study system. Either active filters or passive filters can be used to reduce the harmonics. The former injects an offset harmonic current to compensate the existing harmonics. The latter provides bypassing paths to harmonic flows. The merit of any candidate harmonic compensation scheme should be judged based on its system-wide impact rather than the individual improvement at local busbars. The final scheme can usually be obtained by optimisation, which provides the best compromise among various objectives and constraints. Appropriate weighting factors can be chosen to set up relative performance priorities among all busbars concerned. The objective function of the above optimisation problem is highly multi-modal, discontinuous and noisy, which is hard for traditional search methods to obtain the global optimum. Furthermore, characteristics of MRT systems are time varying and dynamic, which pose additional problems to traditional techniques. The effectiveness of the genetic algorithm (GA) approach for such type of problem has been well documented. Using the GA approach for filter placement, it is guaranteed that the maximum possible distortion level for all the studied cases will not exceed the required limit. In the study MR T system, all the trains are scheduled to satisfy the regularity requirement under all operation modes. The regularity requirement is used to set up the relationship between train locations and the corresponding power consumption. After all the train locations and consumption are available, the contribution from each harmonic source (the rectifiers/inverters) to harmonic distribution is evaluated by harmonic loadflow study. Based on the results from each harmonic loadflow study, the GA approach makes its way to search for the worst case of distribution. Once the worst case is identified, the design and placement of an appropriate filter are treated in a similar manner. Filter parameters and possible placements are part of the GA string. The final solution to the problem provides a reasonable harmonic compensation scheme. Results obtained in various case studies demonstrate the advantages of using the GA approach in solving the harmonic reduction problem.