Research on Online Estimation of Intrinsic Parameters of Three-Phase Harmonic Filter Capacitor

三相諧波濾波器電力電容器在線內部參數估計數校正之研究

Student thesis: Master's Thesis

View graph of relations

Author(s)

Related Research Unit(s)

Detail(s)

Awarding Institution
Supervisors/Advisors
Award date8 Sep 2022

Abstract

Switching devices, non-linear load, are all around our life such as smart phone charger, solar inverter and variable speed motor driver. Therefore, harmonics generation significantly increased in power common coupling (PCC). Passive harmonic filter is essential device to improve the harmonic pollution, which is composed of damping resistor, filter inductor and capacitor. However, the capacitor degrades with the operating environment, working time taken and excessive power injection. Passive components reliability becomes more important in modern power system. This thesis presents a study on capacitor monitoring of passive harmonic filter.

In order to diagnose the condition of capacitors, the value of the capacitance is the key factor which indicates the health of the capacitor. This thesis provides solution for online monitoring the health of capacitor by microcontroller. Modified-particle swarm optimization (m-PSO) is applied for capacitance estimation which based on sensing time series capacitor phase voltages and line currents. Furthermore, online monitoring technique prevent disturbance of harmonic filter normal operation. The traditional way of checking condition of capacitor is based on disconnecting the capacitor under test physically from the harmonic filter first. Online capacitor estimation algorithm without disturbances were proposed to monitor the capacitance degradation of encapsulated delta-connected capacitor. The proposed technique has been verified by a prototype diagnosis device with experimental results.

Besides, non-linear least square method (LSM) is used for comparing with proposed algorithm. Since memory storage, power consumption and number of sensors are critical for IoT device, the performance between two algorithms with minimum resources for Internet of Thing (IoT) application are studied.