MPPT for PV systems using deep reinforcement learning algorithms

Ávila, Luis; Paula, Mariano De; Carlucho, Ignacio; Reinoso, Carlos Sanchez

doi:10.1109/tla.2019.9011547

Cited by 22 publications

(8 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If 𝛼 is equal to 1, it means all of particles equal to 𝑃 𝑏𝑒𝑠𝑡 . Initial condition is determined by utilizing particles' duty cycle and power from the last SO iteration, which expressed by (11) and (12):…”

Section: Proposed Mppt Techniquementioning

confidence: 99%

“…Some researchers have proposed intelligence based MPPT techniques to handle the drawbacks of conventional method. The techniques used are artificial intelligence (AI) [7], fuzzy logic control (FLC) [8], [9], machine learning (ML) [10], and deep learning [11]. Artificial neural network (ANN) works based on its ability to learn from data set input.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Hybrid photovoltaic maximum power point tracking of Seagull optimizer and modified perturb and observe for complex partial shading

Windarko

Sholikhah²,

Habibi³

et al. 2022

IJECE

View full text Add to dashboard Cite

<span>Due to natural randomness, partial shading conditions (PSCs) to photovoltaic (PV) power generation significantly drop the power generation. Metaheuristic based maximum power point tracking (MPPT) can handle PSCs by searching PV panels’ global maximum power point (GMPP). However, trapped at local maxima, sluggishness, continuous power oscillations around GMPP and inaccuracy are the main disadvantages of metaheuristic algorithm. Therefore, the development of algorithm under complex PSCs has been continuously attracting many researchers to yield more satisfying results. In this paper, several algorithms including conventional and metaheuristic are selected for candidate, such as perturb and observe (P&O), firefly (FF), differential evolution (DE), grey wolf optimizer (GWO) and Seagull optimizer (SO). From the preliminary study, SO has shown best performance among other candidates. Then, SO is improved for rapid global optimizer. Modified variable step sizes perturb and observe (MVSPO) is applied to enhance the accuracy tracking of SO. To evaluate the performances, high complexity multipeak partial shading is used to test the algorithms. Statistical results are also provided to analyze the trend of performances. The proposed method performances are shown better fast-tracking time and settling time, high accuracy, higher energy harvesting and low steady-state oscillations than other candidates.</span>

show abstract

Section: Proposed Mppt Techniquementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Hybrid photovoltaic maximum power point tracking of Seagull optimizer and modified perturb and observe for complex partial shading

Windarko

Sholikhah²,

Habibi³

et al. 2022

IJECE

View full text Add to dashboard Cite

show abstract

“…9. The first is the family of brain-inspired computing methods, which comprises artificial neural network (ANN) [32], fuzzy logic control (FLC) [33], machine learning [34], and deep learning [35]. The second is chaos optimisation [36] and the third one is the family of metaheuristic algorithms.…”

Section: Soft Computing Gmpptmentioning

confidence: 99%

Recent developments of MPPT techniques for PV systems under partial shading conditions: a critical review and performance evaluation

Kermadi

Salam

Eltamaly

et al. 2020

IET Renewable Power Generation

View full text Add to dashboard Cite

This review covers global maximum power point tracking (GMPPT) methods for photovoltaic (PV) systems under partial shading conditions. Unlike the previous review works that primarily focused on soft computing and hybrid GMPPT, this study gives exclusive attention to the improvement achieved by the conventional MPPT (perturb and observe, hill climbing, and incremental conductance). The improved methods include the popular 0.8 × V oc model and, more recently, the skipping algorithms. In addition to providing qualitative descriptions of the available techniques, this work also attempts to provide a fair evaluation of GMPPT to determine their comparative performances. The competing algorithms, which are selected to represent every category (conventional and soft computing and hybrid MPPT), are benchmarked under carefully selected operating conditions and shading scenarios. The evaluation is focused on four main criteria: tracking accuracy, convergence time, length of voltage fluctuations, and transient efficiency during the search for the global maximum power point. The results obtained from this study can become a basis for researchers and designers to select the best MPPT technique for their respective applications.

show abstract

“…Generally, 75% of the data is utilized for training, and the remaining 25% for testing the model. Image-based machine learning [14] and reinforcement learning algorithms [15,16] have been applied to PV systems for MPPT. A converter is required to operate the PV panel at the MPP.…”

Section: Introductionmentioning

confidence: 99%

A New Multivariate Linear Regression MPPT Algorithm for Solar PV System with Boost Converter

Mahesh

Meyyappan

Alla³

2022

ECTI-EEC

View full text Add to dashboard Cite

Operating solar photovoltaic (PV) panels at the maximum power point (MPP) is considered to enrich energy conversion efficiency. Each MPP tracking technique (MPPT) has its conversion efficiency and methodology for tracking the MPP. This paper introduces a new method for operating the PV panel at MPP by implementing the multivariate linear regression (MLR) machine learning algorithm. The MLR machine learning model in this study is trained and tested using the data collected from the PV panel specifications. This MLR algorithm can predict the maximum power available at the panel, and the voltage corresponds to this maximum power for specific values of irradiance and temperature. These predicted values help in the calculation of the duty ratio for the boost converter. The MATLAB/SIMULINK results illustrate that, as time progresses, the PV panel is forced to operate at the MPP predicted by the MLR algorithm, yielding a mean efficiency of more than 96% in the steady-state operation of the PV system, even under variable irradiances and temperatures.

show abstract

MPPT for PV systems using deep reinforcement learning algorithms

Cited by 22 publications

References 17 publications

Hybrid photovoltaic maximum power point tracking of Seagull optimizer and modified perturb and observe for complex partial shading

Hybrid photovoltaic maximum power point tracking of Seagull optimizer and modified perturb and observe for complex partial shading

Recent developments of MPPT techniques for PV systems under partial shading conditions: a critical review and performance evaluation

A New Multivariate Linear Regression MPPT Algorithm for Solar PV System with Boost Converter

Contact Info

Product

Resources

About