Photovoltaic shunt active power filter based on indirect self-charging with step size error cancellation and simplified adaptive linear neuron

• Main Author: Mohd Zainuri, Muhammad Ammirrul Atiqi
• Format: Thesis
• Language: English
• Published: 2017
• Online Access: http://psasir.upm.edu.my/id/eprint/71139/1/FK%202017%2034%20-%20IR.pdf; http://psasir.upm.edu.my/id/eprint/71139/
• Institution: Universiti Putra Malaysia

Current harmonics is one of the main power quality problems which can be mitigated by using shunt active power filter (SAPF). Integrating SAPF with photovoltaic (PV), also known as PV SAPF, is among the best option as it provides alternative energy source to operate the SAPF rather than depending on energy from the grid supply and at the same time maintaining Total Harmonics Distortion (THD) below 5%. DC-link capacitor voltage control and harmonics extraction algorithms, are giving high impact to overall SAPF’s performance. In DC-link capacitor voltage control, the existing works on direct self-charging algorithm still have many drawbacks in terms of overshoot, undershoot and response time, especially during dynamic operation. Meanwhile, the existing harmonics extraction algorithm known as modified Widrow- Hoff adaptive linear neuron (ADALINE) algorithm, still has unnecessary features which unfortunately disturbs performance of the algorithm to extract harmonics accurately in both steady-state and dynamic operations. Therefore, this research work proposes design and development of single-phase PV SAPF with a new DC-link capacitor voltage control algorithm named as indirect self- charging with step size error cancellation, and a new harmonics extraction algorithm named as simplified ADALINE. In the indirect self-charging with step size error cancellation, a new technique has been introduced in operation of the self-charging algorithm, known later as indirect control technique. Meanwhile, the simplified ADALINE algorithm has been improved from its existing version by removing cosine component according to symmetrical theory of periodic signal, minimizing large average square error by removing sum of elements, and by modifying weight updating technique leads to introduction of fundamental active current updating technique. In methodology, topology of PV SAPF was designed first, and followed by all control algorithms with special attention to both proposed algorithms. For comparison purpose, the existing DC-link capacitor voltage control and harmonics extraction algorithms were modeled too. Two nonlinear loads, which are inductive and capacitive, and PV source with different level of irradiances were used to test the PV SAPF by focusing on the performances of both proposed algorithms, under steady-state operation. The testing under dynamic operation covers change of nonlinear loads, on-off operations between PV and SAPF, and change of irradiance levels. Laboratory prototype was then developed and digital signal processor (DSP) TMS320F28335 was used to perform the computation of algorithms. Similar tests as in the simulation work were carried out in the laboratory. From both simulation and experimental results, PV SAPF with both proposed algorithms show better performances as compared to the existing algorithms. The indirect self-charging with step size error cancellation performs with high accuracy (99.96 to 100%), low overshoot and undershoot (0.13% to 1%), and fast response time (less than 0.5s). Reduction of energy losses between 36 J to 86 J has been achieved during various dynamic operations of the DC-link capacitor. Meanwhile, the simplified ADALINE performs with lower THD values between 1.5% to 3.24% and high percentages of source power reduction between 4.7% to 23.7% with different nonlinear loads and irradiance levels. In conclusion, PV SAPF with both proposed algorithms have successfully been developed and performed for better improvement of harmonics mitigation and renewable energy utilization.