1 Introduction

According to the United Nations Environment Program (UNEP) briefing note on its Sustainable Development Goal 7 (SDG 7), electricity is unavailable for one billion people in the world while more than three billion people still cook their food and heat their homes using solid fuels like wood. Worse still, four million people die from these wasteful practices and air pollution [1]. To reduce human health problems around the globe, UNEP calls for a reduction of emissions caused by fuel combustion by 40%. Achieving SDG 7 will certainly act to have a positive impact on other SDGs such as combating climate change (SDG 13) and helping to end global poverty (SDG1) by achieving energy justice in developing countries [2, 3]. Among the renewable energy sources (RES), solar energy is the promising alternative and the most useful energy from an ecological point of view as it is an available and a clean energy.

According to the global status report (REN21) [4], “the total installed capacity of RES was about 3146 GW, at the end of 2021” [4]; “hydropower is 1195 GW, PV is 942 GW, wind is 845 GW, bio-power is 143 GW, geothermal is 14.5 GW, concentrating solar thermal power is 6 GW, and ocean power is 0.5 GW” [4]. Even if the largest contribution is from hydropower, PV exposed the fastest growth rate among all RES from 2016 to 2021. The total installed capacity of PV from 2011 to 2021 is presented in Fig. 1 [4, 5].

Fig. 1
figure 1

Total installed capacity of PV along with an annual increment from 2011 to 2021. Used under CCBY. https://www.mdpi.com/1996-1073/16/15/5665 [5]

Full size image

Understanding the technology of solar energy extraction and optimization, especially in developing countries where advances have been seen since 2017, would certainly help to strengthen efforts to achieve 2030 Agenda for the SDGs for all the globe [1]. Advances of solar energy for SDG are described in [6]. Furthermore, “Photovoltaic (PV) systems have developed to be the cheapest source of electrical power in areas with high solar potential, with low cost, 0.01567 US$/kWh in 2020, panel prices have declined by the factor of 10 within a decade” [7]. The environmental impacts of solar energy are told in [8] where a comprehensive review of various applications of solar energy is given in [9].

In general, a PV power system can be either a stand-alone system or grid-connected [10]. In grid-connected PV systems, the energy produced is either consumed on-site or sold to the grid in case of surplus production. When there is a deficit, or during unfavorable moments, the grid supplies the site. Stand-alone PV systems are used in villages and isolated companies in remote areas. They are also helpful for many applications like health, agriculture, and utilities [9]. Solar PV panels absorb sunshine and convert it to electricity. This energy can power devices or be saved in batteries. However, several problems related to low conversion efficiency, high-cost level of PV panels and multiple local peaks of energy caused by partial shading conditions (PSCs) may be met [11]. PSCs are phenomena occurring in PV cells due to the uneven radiation distribution in solar panels. The main goal is minimizing the fluctuations over the maximum power point (MPP) and increasing efficiency and tracking speed under steady-state or rapid changing of climatic conditions. To optimize energy extraction in PV systems, several maximum power point tracking (MPPT) methods are proposed in the literature for uniform solar irradiance conditions (USICs) and for PSCs [11,12,13,14]. The most used techniques are described in [15, 16]. MPPT algorithms are evaluated and classified using different criteria including software and hardware complexity, tracking speed, convergence time and speed, efficiency, accuracy, number of required sensors, cost and so on [17,18,19,20,21].

A “Scopus” survey was conducted in this paper around classification approaches for MPPT approaches evaluation criteria as well used by researchers to compare these approaches. It has been found that a large number of research and survey articles have been published since 1985 for which different classifications are proposed. MPPT techniques are divided into two groups: MPPT techniques for UICs and MPPT techniques for PSCs. A selection method is considered in order to extract not only basic classifications but also the most recent advances related to these approaches. The result of the selection is thirty articles published in the last three years including surveys and research papers able to give an overview of this field of research as well as the latest advances and the latest trends. Different taxonomies of MPTT approaches are found and revised. Compared to the related survey papers, not only this work revises main approaches of MPPT developed so far, an analysis for which a great need exists from time to time, but also it uses a strict minimum number of well selected articles with distinguishable taxonomies chosen from different editors/journals. Additionally, this paper tries to answer the following key questions:

  • What are the most important taxonomies related to MPPT approaches and what are the main performance criteria?

  • What are the current trends in MPPT algorithms?

  • Compared to each other, what are the strengths and weaknesses of the most newly proposed approaches?

  • What are the future directions and challenges?

This work is organized as follows: the following section describes basics on PV systems as well those of MPPT approaches. In Sect. 3, the systematic review with the paper’s selection methodology is introduced. Section 3 depicts the main taxonomies told in selected survey papers. In Sect. 4, the most recent advances in the field are exposed pointing out the strong points of newly proposed approaches and comparative analyses. Finally, Sect. 5, presents future directions and open challenges in the field.

2 Basics on a PV system and MPPT approaches

2.1 Solar PV system with MPPT

It is well recognized that MPPT is an operating point approach connected between PV arrays and a power converter to extract the maximum power energy. To perfect energy extraction in PV systems at any environmental condition, especially solar irradiance, and temperature, MPPT techniques are used. The basic block diagram of a typical PV system with MPPT is shown in Fig. 2.

Fig. 2
figure 2

Solar PV system with MPPT

Full size image

2.2 Partial shading conditions

Due to varying shadows over large surfaces of PV modules, PV cells are constrained to experience nonuniform solar irradiation conditions disturbing the nonlinear characteristics of these systems. This irregular distribution is caused by clouds, bird droppings, and shadows from buildings and trees. An example of such phenomena is illustrated in Fig. 3. Comparable situation directly affects the power–voltage (P–V) and the current–voltage (I–V) nonlinear curves of the PV system as shown by Fig. 4. The I–V and P–V characteristics can show multiple local maximum power points (MLMPP) and only one global maximum power point (GMPP) under PSCs. Such a situation requires an identification of the GMPP for a better extraction of the PV energy. Researchers have proposed and have practically experimented with different MPPT techniques to reach stable optimized outputs from PV systems under PSCs using different approaches. The following section presents the systematic review as well as the selection methodology conducted to find the main taxonomies of these approaches.

Fig. 3
figure 3

Solar PV arrays subject to PSCs caused by clouds and shadows from trees [22]. Used under CCBY. https://doi.org/10.1109/ACCESS.2020.3028609

Full size image
Fig. 4
figure 4

IV–PV characteristic curves of a solar PV array under PSCs a I–V b P–V characteristics [23]. Used under CCBY. https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-rpg.2019.1163

Full size image

3 The systematic review

3.1 Survey methodology

The systematic review, respecting the PRIMA methodology [24], was managed using the “Scopus database” by selecting the keyword “MPPT”. The objective is to answer the fundamental questions already asked in the introduction section. The exploration shows the wealth of the considered research field, and more than 10,408 document results are found in January 2023. The investigation was limited in a second stage to journal papers written in English by excluding publication in 2023. It has been found 3573 journal papers written in English over the period 1985–2022. As shown by Fig. 5, the peak of production for this category is seen in 2019 with 491 publications. Over the total number of these journal papers, more than 60% are written during the last four years (2018–2022).

Fig. 5
figure 5

Data on MPPT techniques literature from Scopus in January 2023 over the period (1985–2022). Query: KEY (mppt) AND (LIMIT-TO (LANGUAGE, "English")) AND (LIMIT-TO (SRCTYPE, "j")) AND (EXCLUDE (PUBYEAR, 2023))

Full size image

We have also investigated the leading countries with the most publications. Figure 6 reveals that the India with 1164 publications (34%), China with 377 (11%), and Algeria with 356 (10%) are the three leading countries in the research production. Furthermore, we can conclude that for the MENA region, the researchers from Algeria, Morocco, Egypt, Saudi Arabia, Iran, and Tunisia are the most active in the field.

Fig. 6
figure 6

Article publications on MPPT techniques by country. KEY (mppt) AND (LIMIT-TO (LANGUAGE, "English")) AND (LIMIT-TO (SRCTYPE, "j")) AND (EXCLUDE (PUBYEAR, 2023))

Full size image

Figure 7 shows the considerable number of review papers published for the period 1991–2022 displaying more than 136 survey papers. The peak of production for this category is seen in 2018, 2020 and 2021 with eighteen survey papers, respectively. Production shows the steepest rising slope since 2011.

Fig. 7
figure 7

Data on MPPT techniques literature from Scopus in March 2022. Query: KEY (mppt) AND (LIMIT-TO (LANGUAGE, "English")) AND (LIMIT-TO (SRCTYPE, "j")) AND (LIMIT-TO (DOCTYPE, "re")) AND (EXCLUDE (PUBYEAR, 2023))

Full size image

3.2 Selection method

The papers selected to build the study consist of a set of thirty articles created as follows: fifteen survey papers published during 2019–2022 and fifteen research papers all published during the two last years. For the latter category, the goal is to find current trends in the field. The complete set (survey and research papers) is built in two stages shown in Fig. 8 according to the PRISMA selection methodology [24]. We suppose that the selected references are able to give an overview of recent advances and new developments in MPPT techniques. For the first set of survey papers, only articles giving distinguish classifications based on evaluation criteria and comparative studies are considered. Using this exclusion criteria leads for example to consider only one survey paper in 2022.

Fig. 8
figure 8

Selection methodology following PRISMA methodology

Full size image

3.3 Selection analysis

Table 1 presents the sample of the fifteen selected survey papers carefully chosen from different publishers/journals and listed in sequential from the earliest to the present. This choice shows that PV MPPT approaches are classified according to different taxonomies using different evaluation criteria. Most selected papers refer to a set of “classical” or “conventional” MPPT approaches widely applied for their simplicity and easy implementation. They are applied for their algorithms lower complexity making them the best techniques for simple applications not requiring high performances. As specified by most survey papers, these methods are known for their efficiency in USICs. However, they show poor dynamic responses and high oscillation dynamics around MMPP under PSCs and rapidly changing weather conditions as they cannot track the GMPP. As shown by Fig. 9, conventional MPPT can be depicted on two groups [14]: online and offline methods. Many selected papers also refer to another category of approaches, the hybrid approaches. A hybrid approach is an advanced approach inspired by the fusion of more than one MPPT approach.

Table 1 Taxonomies of MPPT approaches and related evaluation criteria
Full size table
Fig. 9
figure 9

Conventional MPPT techniques. Adopted from [14]. Used under CCBY. https://ieeexplore.ieee.org/document/9134709

Full size image

3.4 Selection synthesis

For better understanding of MPPT techniques, we explore in this section the different classification approaches considered in the literature and presented in the last section, and we group them in five large categories:

  1. (1)

    Classification depending on the tracking algorithm [16, 23, 27,28,29,30]

  2. (2)

    Classification based on the tracking nature [20, 21, 25]

  3. (3)

    Classification based on selected variables (inputs, parameters…) [14, 26]

  4. (4)

    Hybrid classification approaches [15, 21]

  5. (5)

    Other [11, 14, 31].

To better illustrate these categorizations, we give in Figs. 10, 11, 12, 13, 14, 15 several examples of classifications approaches.

Fig. 10
figure 10

Classification of MPPT approaches on three classes depending on the tracking algorithm: (1) classical algorithms, (2) intelligent algorithms, and (3) optimization algorithms [23]. Used under CCBY. https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-rpg.2019.1163

Full size image
Fig. 11
figure 11

Classification of MPPT approaches on four classes depending on the tracking algorithm: (1) classical algorithms, (2) intelligent algorithms, (3) optimization algorithms, and (4) hybrid Algorithms [16]. Used under CCBY. https://ieeexplore.ieee.org/abstract/document/9171659

Full size image
Fig. 12
figure 12

Classification of soft-computing MPPT approaches in three classes depending on the tracking algorithm: (1) AI, (2) BioI and (3) hybrid algorithms. Adopted from [14] and [28]. Used under CCBY. https://ieeexplore.ieee.org/document/9134709. https://www.mdpi.com/2071-1050/13/19/10575

Full size image
Fig. 13
figure 13

Classification of MPPT approaches based on the tracking nature [20]. Used under CC BY 4.0. https://www.sciencedirect.com/science/article/pii/S2666188820300137

Full size image
Fig. 14
figure 14

Classification of MPPT approaches based on the tracking nature [25]. Adapted and used under CC BY 4.0. https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-rpg.2018.5946. Linear reoriented coordinates method (LRCM); Incremental resistance (INR); Variable step size (VSS); Parasitic capacitance (PC); Ripple correlation control (RCC); Analytical solution (AS); Variable step-size (VSS); Load current or voltage maximization (LCVM); Linear current control (LCC); Constant voltage (CV); Constant current (CC); short circuit current (SCC); Open-circuit voltage (OCV); Temperature parametric (TP); Best fixed voltage (BFV); PV output senseless (POS); sliding control (SC); Extremum seeking method (ESM); Newton–Raphson method (NRM); Secant method (SM); Central point iterative (CPI); False position method (FPM); Brent method (BM); Bisection search method (BSM); fuzzy logic control (FLC); Artificial neural network (ANN); Adaptive neuro-fuzzy inference system (ANFIS); Evolutionary algorithm (EA); firefly algorithm (FA); Predictor method (PM); Chaos Optimization search (COS); Ant colony optimization (ACO); Artificial bee colony (ABC); Shuffled frog leaping algorithm (SFLA); Bayesian network (BN);

Full size image
Fig. 15
figure 15

Categorization of MPPT approaches according to sensed variables: Some examples. Adopted from [26]. Used under CC BY 4.0. https://ieeexplore.ieee.org/document/9212352. Solar irradiance (λ); Variations of PV array’s temperature (T); Sensed PV arrays’ terminal voltage (V); Sensed PV arrays’ terminal current (I)

Full size image

4 New trends on GMPP tracking approaches

To explore contemporary trends around MPPT approaches, Table 2 has been compiled.

Table 2 Recent advances in GMPP tracking approaches and related comparative analysis
Full size table

The analysis of the selected articles shows that trends are moving more towards the use of metaheuristic and swarm algorithms including PSO, GWO, ACO, ABC, CS and so on. Figure 16 shows a summary of current trends in tracking approaches depicted in six groups including the extended class of metaheuristic and BioI algorithms [5]. The later methods have proven high performance to manage tracking problems under different weather scenarios and hardware configurations by reaching GMPP [28, 30]. Table 3 gives a comparative analysis between the most used Metaheuristic methods with high ability to track GMPP under PSCs [44, 46]. Conventional methods like P&O, INC, and HC, used for comparative studies with new proposed methods, are simple but less exact in PSCs. Their responses are slow and present oscillations under PSCs [25]. Indeed, these algorithms are not suitable to be employed in PSCs due to their convergence to local maxima [22]. “The intelligent prediction methods such as FLC, ANN or ANFIS have an ability to manage non-linearities without an exact mathematical model supplies conspicuous tracking efficiency. They have downsides as they are expensive [29].” Furthermore, they need a huge amount of data for their training process which imposes an extreme load on processor memory [22]. Recent proposed swarm algorithms [34, 36, 37, 39, 41, 42] and enhanced ones [22, 32, 33, 35, 40, 44, 45] have shown improved performances compared to earliest ones. Figure 17, 18 give examples of such performance. The contribution of the new swarm algorithms was proved in terms of stability, oscillations around MPP, efficiency, settling time, robustness, dependance to different module configurations and sensitivity to different shading scenarios and dynamically varying insolation conditions.

Fig. 16
figure 16

Summary of current trends in tracking approaches depicted in six groups by extending the class of metaheuristic and BioI algorithms [5]. Used under CCBY. https://www.mdpi.com/199-1073/16/15/5665

Full size image
Table 3 Metaheuristic Algorithms with high ability to track GMPP under PSCs [44, 46]
Full size table
Fig. 17
figure 17

Comparative approach between most used approaches and of the most recently proposed ones, FOA algorithm [43]. Used under CCBY. https://ieeexplore.ieee.org/document/9969612

Full size image
Fig. 18
figure 18

Comparative approach between PSO, Jaya and the improved Ajaya algorithm on convergence time [22]. Used under CCBY. https://ieeexplore.ieee.org/document/9212370

Full size image

5 Future directions and challenges

This section exposes potential future directions and challenges that can be envisioned. The selected works was based on journal articles written in English published in 2023 or still in press using the keyword “MPPT”. By compiling these last publications, we detected the following still relevant open directions and several challenges:

  • Implementing novel metaheuristics approaches for GMPP tracking [47].

  • Enhancing GMPP metaheuristics approaches by improving existing algorithms [48].

  • Proposing new hybrid approaches [5, 49,50,51,52,53]. The objective is to advance the performances of GMPP algorithms by increasing the tracking efficiency and lowering computational burden of hardware [16]. Frequently, combining off-line and online techniques can lead to enhancing the running of the entire system.

  • Proposing new GMPPT approaches using data-driven energy extraction and trained deep learning and/or machine learning models [53, 54].

  • Proposing novel nonlinear robust controllers to optimize energy extraction [55, 56].

  • Proving the validity of the new proposed approaches for different standard benchmarks and using structured methodologies including different scenarios (“zero shading/non-shading, weak partial shading, strong partial shading, and continuously changing weather conditions [44]”), different hardware configurations, a variety of procedures, inputs, and perturbations. Figure 19 gives a graphical representation of such an example of structured methodology used to validate a new controller model for GMPP tracking [55].

  • Proving the validity of the new proposed approaches via comparative approaches using relevant performance criteria including structures, procedures, computations, oscillations, implementation, memory, efficiency, system dependency, tracking speed, and performance under different shading conditions [44]. Figure 20 shows an example of an explicit comparative approach including relevant criteria.

  • Proposing and testing novel MPPT approaches using hybrid energy renewable sources (HERS) combining two or more modes of electricity generation together like PV systems and wind turbines [57] and photovoltaic-thermoelectric generation systems [53]. These technologies often incorporate a storage system (battery, fuel cell) to ensure maximum supply reliability and security [52, 57]. Figure 21 gives two examples of HERS.

  • Proving the applicability of the novel proposed approaches for real applications including power system supplying electric vehicles [56], water heating systems [58], water pumping systems [59] and so on. Fig. 22 shows an example of such applications.

Fig. 19
figure 19

Graphical representation of an example of structured methodology for a proposed GMPP technique including different scenarios, architectures, inputs, and perturbations [55] used under CCBY. https://pcmp.springeropen.com/articles/10.1186/s41601-023-00288-9

Full size image
Fig. 20
figure 20

Example of an explicit comparative approach including the relevant criteria of tracking speed, operating at PSC, oscillation, complexity, and efficiency [55]. Adopted under CCBY. https://pcmp.springeropen.com/articles/10.1186/s41601-023-00288-9. Adaptive robust fuzzy proportional-integral (ARFPI); Adaptive sliding mode MPPT controller with quantized input (QI SMC); Variable step backstepping controller (VS-BS); Improved bat algorithm and fuzzy logic controller (IBA FLC); coarse and fine control algorithm (Coarse and Fine); Load voltage based MPPT (LVB); Reduced oscillation P&O (ROP&O); Steady output and fast tracking MPPT (SOFT MPPT); Fuzzy aided integer order proportional integral derivative with filter (FPINDN); Lyapunov-based robust model reference adaptive controller (LRMRAC)

Full size image
Fig. 21
figure 21

Graphical representation of two cases of HERS: a Wind turbine/PV/pump-hydro storage/biomass system. Used under CCBY. https://www.mdpi.com/1996-1073/14/2/489. b Wind turbine/PV/system associated with an electrochemical/hydraulic storage system. Used under CCBY. https://www.mdpi.com/2079-9292/11/20/3261

Full size image
Fig. 22
figure 22

Example of an application of a novel approach for photovoltaic power system supplying electric vehicle. Adopted under CCBY [56]. https://www.sciencedirect.com/science/article/pii/S2352484723002317

Full size image

6 Conclusion

Since the eighties, researchers around the globe have been working to improve the performance of solar panels. Several MPPT approaches have been proposed to extract the highest amount of power from the PV arrays. Through this survey paper, it is clear that the trends are moving towards artificial intelligence-based approaches. However, it is obvious there is not a 100% guaranteed algorithm to give the best ability in any conditions. Any algorithm should be carefully evaluated using different modules configurations under different shading scenarios. Compared with conventional MPPT techniques, all intelligent MPPT like FLC, ANN and ANFIS techniques show high tracking efficiency of MPP and less steady-state oscillation in rapidly changing weather conditions without prior knowledge of the mathematical model. However, these methods suffer from implementation complexity, long response times, big data processing, and high realization cost. Hybrid MPPT techniques are more efficient, and they are well recommended for complex applications for which PV systems are susceptible to output power fluctuation. They are known for fast convergence, utmost precision, and ability to predict nonlinearities of a PV cell without falling into local MPP under PSCs. Finally, metaheuristic and swarm-based algorithms have proven the highest performances to manage tracking problems under rapidly changing weather conditions and PSC. They have the ability to reduce the computation burden and improve accuracy and convergence speed. They are recommended for complex optimization problems and applications needing high performance. Compared to other MPPT groups, metaheuristic algorithms offer better performance in tracking speed, efficiency accuracy, sampling rate and stability.