Abstract
In order to extract the maximum photovoltaic (PV) power under stochastic weather conditions, a maximum power point tracking (MPPT) technique is required. PV modules behave nonlinearly in unpredictable weather conditions, and the effectiveness of many control strategies under such conditions especially conventional ones—decreases significantly. It is essential to maximize the use of PV power in the system; thus, MPPT algorithms have been developed to ensure that grid integrated PV systems work best at the maximum power point regardless of weather conditions. This research presents a novel HLO-ANN (Horned Lizard optimized artificial neural network) MPPT technique for grid-integrated solar photovoltaic systems which is the improved version of the ANN-MPPT, developed using a novel HLO technique, which effectively optimizes weights of the neural network. The irradiance and temperature datasets from the NASA website have been utilized to train the neural network. The proposed methods have been evaluated by comparing their simulation results to other ANN optimized MPPT techniques: The results showed that the proposed method outperforms the other methods in terms of robustness, DC bus voltage regulation, tracking speed, convergence to the minimum error value, and tracking efficiency in maximizing the harvested power from the PV systems. Furthermore, the EN50530 MPPT efficiency test was carried out during both fast and slow-varying irradiance levels to examine the effectiveness of proposed algorithms.
Introduction
Solar energy has significant impact on preserving the planet and creating a cleaner society. PV modules are becoming cost-effective and more efficient as a result of the rapid growth in technology. However, due to their dependency on meteorological conditions for energy generation, solar photovoltaic systems (SPS) are generally uncertain. Sophisticated control strategies are critical for ensuring solar PV systems operating efficiently. Solar cells have nonlinear I-V and P-V characteristics influenced by external variables such as solar irradiation, humidity, temperature, geographical location, and various diverse dynamic conditions. Therefore, to extract maximum power the SPS operates at maximum power point (MPP) which is the point on the power-voltage curve where the solar cell produces maximum power [1].
Prior research has shown that implementing tracking techniques for MPP may significantly enhance the efficiency of PV systems. A reliable control technique maintains the MPP under all environmental circumstances ensuring that the system functions optimally. Over the last two decades, solar PV systems have employed several maximum power point tracking (MPPT) algorithms to achieve MPP operation for various temperature and irradiance conditions. These MPPT techniques are broadly divided into two categories: conventional MPPT and advanced MPPT techniques [2]. The most well known conventional MPPT techniques are Hill climbing, Perturb and observe (P&O) and Incremental Conductance (INC) these techniques are simple, easy to implement and have low implementation cost and these techniques mainly depend upon the step change of voltage and duty cycles [[3], [4], [5]].. The primary limitation of conventional MPPT techniques have slow tracking speed, significant steady-state oscillation of voltage near the maximum power point. Also, these strategies are ineffective for dealing with sudden fluctuations in solar irradiance and temperature. To overcome these shortcoming advanced MPPT techniques such as sliding mode control (SMC), model predictive control (MPC), fuzzy logic, artificial neural network (ANN), optimization based and hybrid MPPT techniques have been introduced [6]. Sliding mode control is commonly utilized in nonlinear control system design due to its resilience, simplicity, and good dynamic behaviours. SMC MPPT technique have been employed to overcome parametric variations. Though SMC has properties to overcome parametric variations and can be easily implemented, this technique suffers from chattering issue and do not exhibit good dynamics in rapidly changing environmental conditions [7,8]. In [9], the authors proposed model predictive control (MPC) based MPPT technique for PV systems that utilizes a smaller number of sensors, which significantly reduces the hardware cost of the system and enhances the efficiency of the system. MPC-based MPPT systems rely on exact modelling of the solar panel and environmental variables, but tunning the predictive model’s parameters to achieve optimal performance can be challenging necessitating significant testing. [10]. In [11,12] fuzzy logic controller and interval type-2 fuzzy logic control strategies has been utilized by authors for grid connected SPS and for hybrid energy storage system. Presented MPPT technique effectively balances tracking speed with steady-state oscillations. But computation of fuzzy logic operations, including fuzzification, rule evaluation, and defuzzification can add overhead or latency and affect system performance and efficiency.
Inspired by the organic neural networks found in human brains, the ANN system was created. The I-V and P-V nonlinearity relationship of a PV system is trained and evaluated using ANN. With the use of inputs such as temperature, irradiance, voltage, current, and metrological data, an artificial neural network (ANN) continuously learns how to adjust the behaviour of the solar power system in order to maximize power [13].
Reference [14] suggested variable step size ANN MPPT algorithm for partial shading condition of solar PV system has been proposed to improve the performance of PV system. ANN offers numerous benefits, such as its exceptional accuracy in simulating nonlinearity and its ability to deal with challenges without the need for prior knowledge or models. Accurate, standardized, and adequate training data is crucial for ANN to function effectively and without large training errors. Developing an ANN model for a solar system might be challenging due to variances in training and operating variables [15]. To overcome these shortcomings and to maximize power extraction from a PV at a dynamic atmospheric conditions researchers employ numerous optimization-based MPPT algorithms such as particle swarm optimization (PSO) [16], improved PSO [17], grey wolf optimization (GWO) [18], cuckoo search optimization (CSO) [19] artificial bee colony (ABC) [20], harmony search (HS) [21], teacher learning based optimization [22], dwarf mongoose optimizer (DMO) [23] etc. PSO, GWO and CSO based MPPT techniques are simpler, easy to implement require less parameters and mathematics in modelling but these techniques have high risk of premature convergence, where the algorithm may converge to a suboptimal solution before exploring the entire search space thoroughly.
Artificial bee colony (ABC) is relatively resistant to local optima due to its ability to maintain a diversified population of candidate solutions via the scout bees’ exploration activity. But when dealing with complex or high-dimensional MPPT challenges, ABC may experience sluggish convergence. Harmony Search (HS) maintains a memory of the best solutions found so far; this memory consideration improves the optimization algorithm to converge near optimal solution. Although HS has several merits including memory consideration and adaptability it also suffers slow convergence similar to ABC algorithm due to which it may take more iteration to reach near optimal point. While TLBO offers advantages such as global optimization, ease of use, and parameter-free nature, which lessens the need for extensive parameter tunning and makes the algorithm easier to deploy, it also has certain drawbacks such as population size sensitivity and exploration-exploitation balance. The ability of dwarf mongoose’s optimization (DMO) to quickly adapt to changes in their surroundings and their agility is useful for tracking the variable maximum power point of solar panels in response to variations in temperature and irradiation.
The Hybrid MPPT algorithm is essentially derived from combining any other novel techniques with a conventional technique or from combining a novel technique with an intelligence or optimization technique, or by combining intelligent techniques with optimization techniques such as an ANN, fuzzy, GA, PSO, GWO, etc. [2]. Hybrid MPPT techniques is basically employed by researchers to overcome the shortcomings of conventional intelligence and optimization based MPPT techniques. In [24] authors propose a hybrid enhanced leader particle swarm optimization (ELPSO) technique to locate global maximum point zones with the use of a traditional perturb and observe (P&O) algorithm. Since MPPT has intrinsic mutations, ELPSO applied to it performs well in the early stages of identifying the global best leader by investigating global regions; however, P&O quickly reverts back after identifying the global solution space. But P&O has been found to create oscillations in photovoltaic systems around the maximum power point. This feature may cause instability or undesirable performance when paired with ELPSO in certain cases. In [25] A hybrid MPPT technique using iterative learning control (ILC) and the perturb and observe (P&O) algorithm is developed. When the operation point is close to the MPP or there is a small irradiance variation, ILC can deal with the periodic variations and eliminate the steady-state oscillations and errors but adapting the combined ILC and P&O technique to diverse environmental circumstances can be challenging requiring periodic retuning.
Hybrid algorithms for MPPT that combine artificial neural networks (ANN) with optimization approaches eliminate the shortcomings of the preceding research and provide substantial benefits such as higher accuracy, faster convergence, robustness, efficiency, flexibility, and learning capability. These benefits make it a highly successful strategy for several hybrid MPPT techniques, such as ANN-PSO MPPT [26], GA-optimized artificial neural network [27], and HHO-optimized ANN technique [28], which have been developed to address non-linearities and uncertainties in solar PV systems and optimize the performance of solar photovoltaic systems, particularly under dynamic and challenging atmospheric conditions. While GA, PSO, and HHO have the ability to look into a variety of solutions, they may struggle to efficiently search high-dimensional and nonlinear solution spaces. This constraint may limit the ability to discover the global optimum for the ANN model, potentially leading to poor MPPT performance.
Although numerous metaheuristics algorithm has been developed based on number of search agents, problem dimensions and maximum number of iterations count, a better novel swarm-based optimization algorithm inspired by horned lizards reptiles named as Horned Lizard Optimization (HLO) algorithm has been introduced recently. This algorithm imitates crypsis, skin whitening/darkening, splashing blood, and move-to-escape defence mechanisms mathematically and these mimicking behaviour of HLO make them suitable to optimize the complex problems. Furthermore, this strategy balances local and global search across the solution space using exploitation and exploration methods. Maintaining a balance between exploration and exploitation is critical for efficient performance because previously developed optimization techniques struggle with it. The Horned Lizard Optimization technique stands out because of its unique biological inspiration, prey prediction mechanism, and adaptability to a wide range of optimization applications. While various optimization methodologies may have distinct benefits and drawbacks, HLO is a viable strategy for dealing with complex optimization problems, particularly ones where biological inspiration and adaptive search tactics are useful [29]. The aforementioned characteristics of the HLO technique can be implemented into the MPPT algorithm to effectively improve ANN parameters like weights and biases. The accuracy with which the ANN tool estimates PV power is highly dependent on the way the network is trained. In order to address this problem, we came up with the HLO-ANN MPPT algorithm. A robust and efficient MPPT controller has been created for SPS systems by combining the HLO algorithm with ANN, that will maximize energy harvesting under variable environmental conditions.
The novelty of this work is to integrate the metaheuristic Horned Lizard optimization algorithm with the well-known adaptive ANN technique to achieve a better tracking system that harvests the maximum possible power from solar photovoltaic systems. To the best of the authors’ knowledge, the HLO-ANN MPPT technique is being employed for the first time in the MPPT algorithms for grid integrated solar photovoltaic systems to effectively manage the MPPT.