Here, a new control algorithm based on functionallinkartificialneuralnetwork (FLANN) has been designed and studied for shunt compensation. FLANN controller is synonymous to extended trigonometric functional non-li...
详细信息
Here, a new control algorithm based on functionallinkartificialneuralnetwork (FLANN) has been designed and studied for shunt compensation. FLANN controller is synonymous to extended trigonometric functional non-linear expansion and is trained online to achieve mitigation of different power quality (PQ) problems. Increased use of power electronics-based loads has contributed to harmonic pollution in power distribution systems and the PQ issues raised need to be effectively addressed. Several other PQ problems exist at the load end such as poor power factor, unregulated voltage, load unbalancing etc. and these can be effectively controlled using a shunt compensator. The designed FLANN controller is based on individual non-linear functional expansion of the input signal and it is updated online by the adaptive least mean square (LMS) corrective learning algorithm. The controller is designed to predict the weighted active component of load current to generate reference supply current for the grid. The algorithm converges fast and has good transient as well as steady-state response. Simulation as well as experimental results are shown for shunt compensator installed for the single-phase grid-connected system and the performance results are compared with some conventional algorithms usually utilised for compensation.
暂无评论