Simulation and Analyze of Global MPPT Based on Hybrid Classical-ANN with PSO Learning Approach for PV System
This paper discusses the optimization efficiency of the Maximum Power Point Tracking (MPPT) algorithms in solar photovoltaic (PV) systems. In fact, to extract the maximum power, the control of the photovoltaic source must meet the requirements of maximum energy performance and reasonable cost (calculation time and simplicity of implementation). To do this, two new MPPT algorithms based on the Perturb and Observe (P&O) method with a fixed step are proposed. The two suggested controllers are used to improve the drawbacks of the conventional P&O method. Indeed to addressing this problem special techniques derived from artificial intelligence are implemented. The first one uses an Artificial Neural Network (ANN) with Levenberg Marquard (LM) learning algorithm as a neural regulator named (POPI-LMNN). The second one uses an Evolutionary Neural Network (ENN) with a Partical Swarms Optimization (PSO) tuning approach as an evolutionary neural regulator named (POPI-PSONN). The controllers applied to the Boost are used as a powerful impedance adaptation and allows an optimal transfer of energy from the solar panels to the loads. A Simulink model is built to simulate the proposed MPPT methods. To better clarify our contribution, a comparative study with the P&O technique is carried out. The simulation results confirm the advantageous contribution of the new “POPI-PSONN”algorithm during sudden changes in solar illumination characterized by its simplicity, its speed of search of the MPP and its independence of the solar panel parameters, as well as its ability to eliminate oscillations. Furthermore, the combination of learning neural networks with PSO solves the problem of convergence towards a global optimum all this gives it very high reliability.
KeywordsPhotovoltaic systems P&O Artificial Neural Network Levenberg Marquard Evolutionary Neural Network Partical Swarms Optimization
- 9.Pareja, M.: Pv cell simulation with Qucs, a generic model of pv cell 20(07) (2013)Google Scholar
- 10.Chtouki, I., Wira, P., Zazi, M.: Comparison of several neural network perturb and observe MPPT methods for photovoltaic applications. In: The 19th International Conference on Industrial Technology (ICIT 2018) lyon, France, pp. 1–6 (2018)Google Scholar
- 12.Bishop, C.M.: Neural Networks for Pattern Recognition. Clarendon Press (1995)Google Scholar
- 14.Aljarah, I., Faris, H., Mirjalili, S.: Optimizing connection weights in neural networks using the whale optimization algorithm. Soft Comput. 18(5) (2017) Google Scholar
- 15.Gürüler, H., Peker, M., Baysa, Ö.: Soft computing model on genetic diversity and pathotype differentiation A novel approach. Electronic J. Biotechnol. 22(1) (2015)Google Scholar