Improving the performance of the MSOGI FLL method in network frequency detection under short circuit conditions

Master's Thesis

in the field of electrical engineering with a focus on power

Abstract

Today, the discussion of system control and monitoring is one of the most important discussions in the power network, and we need to know the exact state of the power network in order to design control devices. By knowing the status of the network at any moment, you can be aware of changes in voltage, current, and frequency, and you can prevent these parameters from leaving the working point with proper performance. One of the parameters that must be recognized is the network frequency, which is discussed in this thesis. The proposed methods for frequency detection based on phase-locked loops, such as the primary phase-locked loop, PLL1 (phase-locked loop) with a simultaneous reference frame, PLL with two separate simultaneous reference frames, and PLL with two second-order generalized integrators are common methods. How to get from the phase-locked loop to the frequency-locked loop (FLL2) and the methods based on the frequency-locked loop include FLL with two second-order generalized trap integrals and FLL with several second-order generalized trap integrals. The mentioned methods have evolved and each method is more efficient than the previous method. Finally, in this thesis, a method to improve the performance of the FLL method with generalized second-order multiple integrators in network frequency detection under short-circuit conditions is presented, and its efficiency is compared and analyzed with conventional methods. 

Foreword

       To ensure the existence of a pure sinusoidal current or voltage signal or in other words a better quality of power, the point that is considered is that the power system should operate without significant loss around its performance. The performance of the system worsens due to voltage fluctuations, frequency changes and waveform changes including harmonics and inter-harmonics [1]. Therefore, in order to improve the power quality (topics related to the discussion of frequency in power quality include harmonics in the power system) fast and accurate estimation of the frequency of the supply voltage is required for an integrated power system [2]. In a power system due to the presence of noise and higher harmonics, the accurate and fast estimation of the frequency may be damaged and disturbed. Due to the power mismatch between production and load demand, a change in the frequency of the system occurs compared to its normal value, and restoring the frequency to its initial value requires calling a corrective operation. [3-4]. The problem of frequency estimation is considered using a number of numerical methods such as Newton-Raphson, weighted least squares and FIR3 through discrete samples of the system voltage Signal processing techniques such as discrete Fourier transforms, least squares error technique, Kalman filter and adaptive notch filters are used to estimate the frequency of noisy signals in the power system. A large number of numerical techniques based on the expansion of the measurement range, including the least square error method, Cramer-Rao bounds method, and maximum likelihood estimation, as well as their practical implementations, have been reported in articles and written texts. However, these methods provide incorrect results due to the presence of noise and harmonics and other variable conditions of the system such as the change in the fault starting angle and the change in the fault resistance. As a result, getting incorrect results in the mentioned conditions is one of the weaknesses of these methods [5]. As a result, a review of new methods such as PLL blocks and its evolution, which respectively includes the primary phase-locked loop, the PLL (phase-locked loop) with a simultaneous reference frame, the PLL with two simultaneous reference frames, Finite Impulse Response, and the second-order generalized wind-locked PLL. It is important. Then how to obtain the FLL block can be described. The goal of all researchers is to obtain the frequency in the shortest possible time with high accuracy in different network conditions. Among the various conditions of the network, the existence of harmonics, imbalance and noise can be mentioned as an example.

The works that have been done so far to detect the frequency of the network in the power system are methods based on frequency locking loops, which for networks that have imbalance, the FLL method with two generalized second-order integrators can detect the frequency well [6]. But it is not able to accurately estimate the harmonic components of the voltage. Therefore, if the network has harmonics, a method should be provided that is able to detect harmonic components as well. In a more recent method, a multiple detector system is used to extract the harmonic components of frequency in power systems, and for this reason it is called FLL with multiple generalized second-order integrators [4]. As a result, the FLL method with multiple generalized second-order integrators can provide a suitable detection of frequency when the network has harmonics and imbalance. According to the simulation results of the articles related to this method, which will be discussed in the next chapter, it shows that the frequency detection time in harmonic and unbalance conditions is 100 milliseconds, which takes five sinus cycles in terms of the power system [4]. SOGI 4 shows the characteristics of a low-pass filter, and for this reason, if the line voltage is distorted, high-order harmonics are prevented from passing, but low-order harmonics such as third harmonics and Fifth, because they are close to the frequency of the network, they pass through the filter and cause the distortion to remain in the main component [6]. The FLL method with several second-order generalized integrals is the last method presented for frequency detection, in which with the help of several SOGI blocks, we will be able to accurately detect sequence components even under severe distortions. Today, frequency detection time is one of the most important parameters in measurement systems. The lower detection time we have, we will be able to perform better control so that the frequency returns to the initial value that is in the desired network [7]. 1-2- Achievements of the thesis In this thesis, the simulation of the FLL method with several generalized second-order integrals is considered and the initial responses are analyzed. In these studies and reviews, it was observed that if the frequency jump of the gamma parameter (one of the adjustable coefficients that determines the detection speed and was considered constant in previous works) is increased in order to follow the changes more quickly and then return to its initial value when the gamma frequency stabilizes, the frequency detection time can be reduced. Analyzes and seeing the effect of changing the parameter K (damping coefficient in frequency detection, which was considered as a unit in previous works) on the time and accuracy of frequency detection, it was observed that if the K coefficient is set for the detection of the main component and harmonic components separately, the damping time can be reduced and the accuracy can be increased. As a result, we first adjust the amount of gamma changes during the frequency jump and choose the best mode, and then by harmonizing the K coefficient of the main component and harmonic components, the detection time is reduced and the accuracy remains at the optimal level. After analyzing the answers and finding the best method, which is FLL with multiple generalized second-order integrals, it is chosen and it will be dealt with more accurately in the third chapter. In the fifth chapter, the improved method is simulated and a detailed explanation of the simulation is given and the results of the simulation are analyzed. Finally, the achievements of the thesis are mentioned in the final chapter and suggestions for future research are given. Knowing the status of the network at any moment can change the voltage and frequency can be notified and the proper functioning of the abatement work out the parameters of the point.