Deep CNN–LSTM-Based DSTATCOM for Power Quality Enhancement in Microgrid

Abstract

Power quality issues and their effective mitigation invariably play a crucial role in a microgrid system. Such power quality problems are often resolved by employing multiple power electronics-based components in the utility grid. This paper is focused on the optimal enhancement of power quality under islanded mode of operation in a microgrid, with a deep Convolutional Neural Network (CNN) with Long Short-term Memory (LSTM) algorithm using distribution static compensator (DSTATCOM). The objective of the research is centered on the reactive power control in DSTATCOM using deep CNN with LSTM for voltage enhancement, minimization of current distortion and reduction of harmonics on a microgrid. This objective can be achieved by the proposed Simulink design model of DSTATCOM intended for improving the power quality in a microgrid. The renewable energy-based power compensator is used for an enhanced and effective control strategy like voltage and current control of the microgrid circuit and uses LSTM-based deep CNN for achieving superior time consumption indicators. Due to varying loads in the microgrid, the reactive power and harmonic voltage and current may be distorted. This problem can be rectified by controlling the microgrid using the LSTM-based deep CNN. This approach consequently reduces the negative-sequence frequency range with the aid of this filtering method in the proposed microgrid circuit. The microgrid is thereafter subjected to different testing conditions and the corresponding simulation results are discussed in relation to existing approaches. The proposed framework was observed to have successfully accomplished harmonic substance and voltage profile enhancement.

Analisis Harmonisa Pada Trainer Lab Generator Sinkron Tiga Fasa NV7017

“…Untuk mengatasi hal ini, arus harmonisa triplen akan bersirkulasi dalam belitan primer delta yang dapat menyebabkan panas berlebih dan kegagalan transformator [16]. Kapasitor daya yang terhubung dengan Wye dapat rusak oleh arus harmonisa triplen yang tinggi selama resonansi paralel [17]. Dalam beberapa kasus, arus harmonisa triplen telah menyebabkan kegagalan pada tahanan pentanahan netral (NGR) [18].…”

Initial condition based real time classification of power quality disturbance using deep convolution neural network with bidirectional long short‐term memory

“…The lager range of raw data are made processed in the proposed DCNN based Bi-LSTM layers using the mathematical modelling of PQDs. The parametric expression of 16 ranges of PQDs signals [72] are categorized as training set and testing set data for detailed training process. The modification of parameters are carried out based on the recommendation of IEEE-1159 standards (as addressed in Table 1).…”

Unified power quality control based microgrid for power quality enhancement using various controlling techniques

“…Then, the reference current control the signals from the transform and the controlled signals are given to the hysteresis controller with the shunt current (I sh(abc) ). And the pulse is given to the shunt APF and the calculation is shown in (17).…”

“…The power compensator is employed for an improved and more efficient control strategy, such as voltage and current regulation of the microgrid circuit, and it employs a convolutional neural network (CNN) having long short-term memory to achieve superior time consumption indicators [17], [18]. The use of the reweighted zero attracting technique, which exhibits good steady-state and transient performance.…”